Datasets:

bigcode

/

jupyter-parsed

like 4

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "## Dimensionality Reduction", "_____no_output_____" ] ], [ [ "from sklearn.decomposition import PCA", "_____no_output_____" ] ], [ [ "### Principal Components Analysis", "_____no_output_____" ] ], [ [ "o_dir = os.path.join('outputs','pca')\nif os.path.isdir(o_dir) is not True:\n print(\"Creating '{0}' directory.\".format(o_dir))\n os.mkdir(o_dir)", "_____no_output_____" ], [ "pca = PCA() # Use all Principal Components\npca.fit(scdf) # Train model on all data\npcdf = pd.DataFrame(pca.transform(scdf)) # Transform data using model\n\nfor i in range(0,21):\n print(\"Amount of explained variance for component {0} is: {1:6.2f}%\".format(i, pca.explained_variance_ratio_[i]*100))\n\nprint(\"The amount of explained variance of the SES score using each component is...\")\nsns.lineplot(x=list(range(1,len(pca.explained_variance_ratio_)+1)), y=pca.explained_variance_ratio_)", "_____no_output_____" ], [ "pca = PCA(n_components=11)\npca.fit(scdf)\nscores = pd.DataFrame(pca.transform(scdf), index=scdf.index)\nscores.to_csv(os.path.join(o_dir,'Scores.csv.gz'), compression='gzip', index=True)\n\n# Adapted from https://stackoverflow.com/questions/22984335/recovering-features-names-of-explained-variance-ratio-in-pca-with-sklearn\ni = np.identity(scdf.shape[1]) # identity matrix\n\ncoef = pca.transform(i)\n\nloadings = pd.DataFrame(coef, index=scdf.columns)\nloadings.to_csv(os.path.join(o_dir,'Loadings.csv.gz'), compression='gzip', index=True)", "_____no_output_____" ], [ "print(scores.shape)\nscores.sample(5, random_state=42)", "_____no_output_____" ], [ "print(loadings.shape)\nloadings.sample(5, random_state=42)", "_____no_output_____" ], [ "odf = pd.DataFrame(columns=['Variable','Component Loading','Score'])\nfor i in range(0,len(loadings.index)):\n row = loadings.iloc[i,:]\n for c in list(loadings.columns.values):\n d = {'Variable':loadings.index[i], 'Component Loading':c, 'Score':row[c]}\n odf = odf.append(d, ignore_index=True)\n\ng = sns.FacetGrid(odf, col=\"Variable\", col_wrap=4, height=3, aspect=2.0, margin_titles=True, sharey=True)\ng = g.map(plt.plot, \"Component Loading\", \"Score\", marker=\".\")", "_____no_output_____" ] ], [ [ "### What Have We Done?", "_____no_output_____" ] ], [ [ "sns.set_style('white')\nsns.jointplot(data=scores, x=0, y=1, kind='hex', height=8, ratio=8)", "_____no_output_____" ] ], [ [ "#### Create an Output Directory and Load the Data", "_____no_output_____" ] ], [ [ "o_dir = os.path.join('outputs','clusters-pca')\nif os.path.isdir(o_dir) is not True:\n print(\"Creating '{0}' directory.\".format(o_dir))\n os.mkdir(o_dir)", "_____no_output_____" ], [ "score_df = pd.read_csv(os.path.join('outputs','pca','Scores.csv.gz'))\nscore_df.rename(columns={'Unnamed: 0':'lsoacd'}, inplace=True)\nscore_df.set_index('lsoacd', inplace=True)\n\n# Ensures that df is initialised but original scores remain accessible\ndf = score_df.copy(deep=True)\n\nscore_df.describe()", "_____no_output_____" ], [ "score_df.sample(3, random_state=42)", "_____no_output_____" ] ], [ [ "#### Rescale the Loaded Data\n\nWe need this so that differences in the component scores don't cause the clustering algorithms to focus only on the 1st component.", "_____no_output_____" ] ], [ [ "scaler = preprocessing.MinMaxScaler()\n\ndf[df.columns] = scaler.fit_transform(df[df.columns])\n\ndf.describe()", "_____no_output_____" ], [ "df.sample(3, random_state=42)", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import os\nimport numpy as np", "_____no_output_____" ], [ "path = \"/users/gpu/utkrsh/data/PennDataset_detections/\"", "_____no_output_____" ], [ "directories = os.listdir(path)", "_____no_output_____" ], [ "d = []\nfor directory in directories:\n files = os.listdir(path+directory+\"/\")\n for file in files:\n try:\n var = np.load(file)\n except OSError:\n if directory not in d:\n d+= [directory]", "_____no_output_____" ], [ "d", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "### The model\n\n$u(c) = log(c)$ utility function \n$y = 1$ Deterministic income \n$p(r = 0.02) = 0.5$ \n$p(r = -0.02) = 0.5$ ", "_____no_output_____" ], [ "### value iteration ", "_____no_output_____" ] ], [ [ "# infinite horizon MDP problem\n%pylab inline\nimport numpy as np\nfrom scipy.optimize import minimize\ndef u(c):\n return np.log(c)", "Populating the interactive namespace from numpy and matplotlib\n" ], [ "# discounting factor \nbeta = 0.95\n# wealth level\nw_low = 0 \nw_high = 10\n# interest rate\nr = 0.02\n# deterministic income\ny = 1\n# good state and bad state economy with equal probability 0.5\n# with good investment return 0.05 or bad investment return -0.05\nws = np.linspace(0.001,10**(1/2),100)**2\nVs = np.zeros(100)\nCs = np.zeros(100)", "_____no_output_____" ], [ "# Value iteration\nfor j in range(50):\n if j % 10 == 0:\n print(j)\n for i in range(len(ws)):\n w = ws[i]\n def obj(c):\n return -(u(c) + beta*(np.interp((y+w-c)*(1+r), ws, Vs) + np.interp((y+w-c)*(1-r), ws, Vs))/2)\n bounds = [(0.0001, y+w-0.0001)]\n res = minimize(obj, 0.0001, method='SLSQP', bounds=bounds)\n Cs[i] = res.x[0]\n Vs[i] = -res.fun", "0\n10\n20\n30\n40\n" ], [ "plt.plot(ws,Vs)", "_____no_output_____" ], [ "plt.plot(ws,Cs)", "_____no_output_____" ] ], [ [ "### policy gradient\nAssume the policy form $\\theta = (a,b,c, \\sigma)$, then $\\pi_\\theta$ ~ $N(log(ax+b)+c, \\sigma)$\n\nAssume the initial value $a = 1$, $b = 1$, $c = 1$, $\\sigma = 1$ \n\n\n$$\\theta_{k+1} = \\theta_{k} + \\alpha \\nabla_\\theta V(\\pi_\\theta)|\\theta_k$$", "_____no_output_____" ] ], [ [ "# simulation step T = 100\nT = 10\ndef mu(theta, w):\n return np.log(theta[0] * w + theta[1]) + theta[2] \n\ndef simSinglePath(theta):\n wPath = np.zeros(T)\n aPath = np.zeros(T)\n rPath = np.zeros(T)\n w = np.random.choice(ws)\n for t in range(T):\n c = np.random.normal(mu(theta, w), theta[3])\n while c < 0.0001 or c > w+y-0.0001:\n c = np.random.normal(mu(theta, w), theta[3])\n wPath[t] = w\n aPath[t] = c\n rPath[t] = np.log(c)*(beta**t)\n if np.random.uniform(0,1) > 0.5:\n w = (w+y-c) * (1+r)\n else:\n w = (w+y-c) * (1-r)\n return wPath, aPath, rPath\n\n\ndef gradientV(theta, D = 100):\n '''\n D is the sample size\n '''\n grad = np.zeros(len(theta))\n newGrad = np.zeros(len(theta))\n for d in range(D):\n wp, ap, rp = simSinglePath(theta) \n newGrad[0] = np.sum((ap - mu(theta, wp))/(theta[3]**2)*(w/(theta[0]*w + theta[1])))\n newGrad[1] = np.sum((ap - mu(theta, wp))/(theta[3]**2)*(1/(theta[0]*w + theta[1])))\n newGrad[2] = np.sum((ap - mu(theta, wp))/(theta[3]**2))\n #newGrad[3] = np.sum((((ap - mu(theta, wp))**2 - theta[3]**2)/(theta[3]**3)))\n grad += newGrad * np.sum(rp)\n grad /= D\n grad[-1] = 0\n return grad\n\ndef updateTheta(theta):\n theta = theta + alpha * gradientV(theta)\n return theta \n\n\nimport time\ndef plot(theta):\n def f(x):\n return np.log(theta[0]*x + theta[1]) + theta[2]\n plt.plot(ws, Cs, 'b')\n plt.plot(ws, f(ws), 'r')", "_____no_output_____" ], [ "# c < 0 or c > w + 5, then reward -100\n# initial theta \ntheta = [1,1,1,0.1]\n# gradient ascend step size \nalpha = 0.001\n# store theta \nTHETA = np.zeros((3,10000))\nfor i in range(10000):\n theta = updateTheta(theta)\n THETA[:,i] = theta[:3]\n plot(theta)", "_____no_output_____" ], [ "theta = [0.4, 1.00560229, 0.74852663, 0.1 ]", "_____no_output_____" ], [ "plt.plot(THETA[0,:])", "_____no_output_____" ], [ "plt.plot(THETA[1,:])", "_____no_output_____" ], [ "plt.plot(THETA[2,:])", "_____no_output_____" ], [ "def V(theta, w, D = 100):\n def sPath(theta, w):\n wPath = np.zeros(T)\n aPath = np.zeros(T)\n rPath = np.zeros(T)\n for t in range(T):\n c = np.random.normal(mu(theta, w), theta[3])\n while c < 0.0001 or c > w+y-0.0001:\n c = np.random.normal(mu(theta, w), theta[3])\n wPath[t] = w\n aPath[t] = c\n rPath[t] = np.log(c)*(beta**t)\n if np.random.uniform(0,1) > 0.5:\n w = (w+y-c) * (1+r)\n else:\n w = (w+y-c) * (1-r)\n return wPath, aPath, rPath\n value = 0\n for d in range(D):\n _,_,rp = sPath(theta,w)\n value += np.sum(rp)\n return value/D", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "##### Copyright 2019 The TensorFlow Authors.", "_____no_output_____" ] ], [ [ "#@title Licensed under the Apache License, Version 2.0 (the \"License\");\n# you may not use this file except in compliance with the License.\n# You may obtain a copy of the License at\n#\n# https://www.apache.org/licenses/LICENSE-2.0\n#\n# Unless required by applicable law or agreed to in writing, software\n# distributed under the License is distributed on an \"AS IS\" BASIS,\n# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n# See the License for the specific language governing permissions and\n# limitations under the License.", "_____no_output_____" ] ], [ [ "# TFRecord and tf.Example\n\n<table class=\"tfo-notebook-buttons\" align=\"left\">\n <td>\n <a target=\"_blank\" href=\"https://www.tensorflow.org/tutorials/load_data/tfrecord\"><img src=\"https://www.tensorflow.org/images/tf_logo_32px.png\" />View on TensorFlow.org</a>\n </td>\n <td>\n <a target=\"_blank\" href=\"https://colab.research.google.com/github/tensorflow/docs/blob/master/site/en/tutorials/load_data/tfrecord.ipynb\"><img src=\"https://www.tensorflow.org/images/colab_logo_32px.png\" />Run in Google Colab</a>\n </td>\n <td>\n <a target=\"_blank\" href=\"https://github.com/tensorflow/docs/blob/master/site/en/tutorials/load_data/tfrecord.ipynb\"><img src=\"https://www.tensorflow.org/images/GitHub-Mark-32px.png\" />View source on GitHub</a>\n </td>\n <td>\n <a href=\"https://storage.googleapis.com/tensorflow_docs/docs/site/en/tutorials/load_data/tfrecord.ipynb\"><img src=\"https://www.tensorflow.org/images/download_logo_32px.png\" />Download notebook</a>\n </td>\n</table>", "_____no_output_____" ], [ "To read data efficiently it can be helpful to serialize your data and store it in a set of files (100-200MB each) that can each be read linearly. This is especially true if the data is being streamed over a network. This can also be useful for caching any data-preprocessing.\n\nThe TFRecord format is a simple format for storing a sequence of binary records.\n\n[Protocol buffers](https://developers.google.com/protocol-buffers/) are a cross-platform, cross-language library for efficient serialization of structured data.\n\nProtocol messages are defined by `.proto` files, these are often the easiest way to understand a message type.\n\nThe `tf.Example` message (or protobuf) is a flexible message type that represents a `{\"string\": value}` mapping. It is designed for use with TensorFlow and is used throughout the higher-level APIs such as [TFX](https://www.tensorflow.org/tfx/).", "_____no_output_____" ], [ "This notebook will demonstrate how to create, parse, and use the `tf.Example` message, and then serialize, write, and read `tf.Example` messages to and from `.tfrecord` files.\n\nNote: While useful, these structures are optional. There is no need to convert existing code to use TFRecords, unless you are using [`tf.data`](https://www.tensorflow.org/guide/datasets) and reading data is still the bottleneck to training. See [Data Input Pipeline Performance](https://www.tensorflow.org/guide/performance/datasets) for dataset performance tips.", "_____no_output_____" ], [ "## Setup", "_____no_output_____" ] ], [ [ "!pip install tf-nightly\nimport tensorflow as tf\n\nimport numpy as np\nimport IPython.display as display", "_____no_output_____" ] ], [ [ "## `tf.Example`", "_____no_output_____" ], [ "### Data types for `tf.Example`", "_____no_output_____" ], [ "Fundamentally, a `tf.Example` is a `{\"string\": tf.train.Feature}` mapping.\n\nThe `tf.train.Feature` message type can accept one of the following three types (See the [`.proto` file](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/core/example/feature.proto) for reference). Most other generic types can be coerced into one of these:\n\n1. `tf.train.BytesList` (the following types can be coerced)\n\n - `string`\n - `byte`\n\n1. `tf.train.FloatList` (the following types can be coerced)\n\n - `float` (`float32`)\n - `double` (`float64`)\n\n1. `tf.train.Int64List` (the following types can be coerced)\n\n - `bool`\n - `enum`\n - `int32`\n - `uint32`\n - `int64`\n - `uint64`", "_____no_output_____" ], [ "In order to convert a standard TensorFlow type to a `tf.Example`-compatible `tf.train.Feature`, you can use the shortcut functions below. Note that each function takes a scalar input value and returns a `tf.train.Feature` containing one of the three `list` types above:", "_____no_output_____" ] ], [ [ "# The following functions can be used to convert a value to a type compatible\n# with tf.Example.\n\ndef _bytes_feature(value):\n \"\"\"Returns a bytes_list from a string / byte.\"\"\"\n if isinstance(value, type(tf.constant(0))):\n value = value.numpy() # BytesList won't unpack a string from an EagerTensor.\n return tf.train.Feature(bytes_list=tf.train.BytesList(value=[value]))\n\ndef _float_feature(value):\n \"\"\"Returns a float_list from a float / double.\"\"\"\n return tf.train.Feature(float_list=tf.train.FloatList(value=[value]))\n\ndef _int64_feature(value):\n \"\"\"Returns an int64_list from a bool / enum / int / uint.\"\"\"\n return tf.train.Feature(int64_list=tf.train.Int64List(value=[value]))", "_____no_output_____" ] ], [ [ "Note: To stay simple, this example only uses scalar inputs. The simplest way to handle non-scalar features is to use `tf.serialize_tensor` to convert tensors to binary-strings. Strings are scalars in tensorflow. Use `tf.parse_tensor` to convert the binary-string back to a tensor.", "_____no_output_____" ], [ "Below are some examples of how these functions work. Note the varying input types and the standardized output types. If the input type for a function does not match one of the coercible types stated above, the function will raise an exception (e.g. `_int64_feature(1.0)` will error out, since `1.0` is a float, so should be used with the `_float_feature` function instead):", "_____no_output_____" ] ], [ [ "print(_bytes_feature(b'test_string'))\nprint(_bytes_feature(u'test_bytes'.encode('utf-8')))\n\nprint(_float_feature(np.exp(1)))\n\nprint(_int64_feature(True))\nprint(_int64_feature(1))", "_____no_output_____" ] ], [ [ "All proto messages can be serialized to a binary-string using the `.SerializeToString` method:", "_____no_output_____" ] ], [ [ "feature = _float_feature(np.exp(1))\n\nfeature.SerializeToString()", "_____no_output_____" ] ], [ [ "### Creating a `tf.Example` message", "_____no_output_____" ], [ "Suppose you want to create a `tf.Example` message from existing data. In practice, the dataset may come from anywhere, but the procedure of creating the `tf.Example` message from a single observation will be the same:\n\n1. Within each observation, each value needs to be converted to a `tf.train.Feature` containing one of the 3 compatible types, using one of the functions above.\n\n1. You create a map (dictionary) from the feature name string to the encoded feature value produced in #1.\n\n1. The map produced in step 2 is converted to a [`Features` message](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/core/example/feature.proto#L85).", "_____no_output_____" ], [ "In this notebook, you will create a dataset using NumPy.\n\nThis dataset will have 4 features:\n\n* a boolean feature, `False` or `True` with equal probability\n* an integer feature uniformly randomly chosen from `[0, 5]`\n* a string feature generated from a string table by using the integer feature as an index\n* a float feature from a standard normal distribution\n\nConsider a sample consisting of 10,000 independently and identically distributed observations from each of the above distributions:", "_____no_output_____" ] ], [ [ "# The number of observations in the dataset.\nn_observations = int(1e4)\n\n# Boolean feature, encoded as False or True.\nfeature0 = np.random.choice([False, True], n_observations)\n\n# Integer feature, random from 0 to 4.\nfeature1 = np.random.randint(0, 5, n_observations)\n\n# String feature\nstrings = np.array([b'cat', b'dog', b'chicken', b'horse', b'goat'])\nfeature2 = strings[feature1]\n\n# Float feature, from a standard normal distribution\nfeature3 = np.random.randn(n_observations)", "_____no_output_____" ] ], [ [ "Each of these features can be coerced into a `tf.Example`-compatible type using one of `_bytes_feature`, `_float_feature`, `_int64_feature`. You can then create a `tf.Example` message from these encoded features:", "_____no_output_____" ] ], [ [ "def serialize_example(feature0, feature1, feature2, feature3):\n \"\"\"\n Creates a tf.Example message ready to be written to a file.\n \"\"\"\n # Create a dictionary mapping the feature name to the tf.Example-compatible\n # data type.\n feature = {\n 'feature0': _int64_feature(feature0),\n 'feature1': _int64_feature(feature1),\n 'feature2': _bytes_feature(feature2),\n 'feature3': _float_feature(feature3),\n }\n\n # Create a Features message using tf.train.Example.\n\n example_proto = tf.train.Example(features=tf.train.Features(feature=feature))\n return example_proto.SerializeToString()", "_____no_output_____" ] ], [ [ "For example, suppose you have a single observation from the dataset, `[False, 4, bytes('goat'), 0.9876]`. You can create and print the `tf.Example` message for this observation using `create_message()`. Each single observation will be written as a `Features` message as per the above. Note that the `tf.Example` [message](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/core/example/example.proto#L88) is just a wrapper around the `Features` message:", "_____no_output_____" ] ], [ [ "# This is an example observation from the dataset.\n\nexample_observation = []\n\nserialized_example = serialize_example(False, 4, b'goat', 0.9876)\nserialized_example", "_____no_output_____" ] ], [ [ "To decode the message use the `tf.train.Example.FromString` method.", "_____no_output_____" ] ], [ [ "example_proto = tf.train.Example.FromString(serialized_example)\nexample_proto", "_____no_output_____" ] ], [ [ "## TFRecords format details\n\nA TFRecord file contains a sequence of records. The file can only be read sequentially.\n\nEach record contains a byte-string, for the data-payload, plus the data-length, and CRC32C (32-bit CRC using the Castagnoli polynomial) hashes for integrity checking.\n\nEach record is stored in the following formats:\n\n uint64 length\n uint32 masked_crc32_of_length\n byte data[length]\n uint32 masked_crc32_of_data\n\nThe records are concatenated together to produce the file. CRCs are\n[described here](https://en.wikipedia.org/wiki/Cyclic_redundancy_check), and\nthe mask of a CRC is:\n\n masked_crc = ((crc >> 15) | (crc << 17)) + 0xa282ead8ul\n\nNote: There is no requirement to use `tf.Example` in TFRecord files. `tf.Example` is just a method of serializing dictionaries to byte-strings. Lines of text, encoded image data, or serialized tensors (using `tf.io.serialize_tensor`, and\n`tf.io.parse_tensor` when loading). See the `tf.io` module for more options.", "_____no_output_____" ], [ "## TFRecord files using `tf.data`", "_____no_output_____" ], [ "The `tf.data` module also provides tools for reading and writing data in TensorFlow.", "_____no_output_____" ], [ "### Writing a TFRecord file\n\nThe easiest way to get the data into a dataset is to use the `from_tensor_slices` method.\n\nApplied to an array, it returns a dataset of scalars:", "_____no_output_____" ] ], [ [ "tf.data.Dataset.from_tensor_slices(feature1)", "_____no_output_____" ] ], [ [ "Applied to a tuple of arrays, it returns a dataset of tuples:", "_____no_output_____" ] ], [ [ "features_dataset = tf.data.Dataset.from_tensor_slices((feature0, feature1, feature2, feature3))\nfeatures_dataset", "_____no_output_____" ], [ "# Use `take(1)` to only pull one example from the dataset.\nfor f0,f1,f2,f3 in features_dataset.take(1):\n print(f0)\n print(f1)\n print(f2)\n print(f3)", "_____no_output_____" ] ], [ [ "Use the `tf.data.Dataset.map` method to apply a function to each element of a `Dataset`.\n\nThe mapped function must operate in TensorFlow graph mode—it must operate on and return `tf.Tensors`. A non-tensor function, like `serialize_example`, can be wrapped with `tf.py_function` to make it compatible.\n\nUsing `tf.py_function` requires to specify the shape and type information that is otherwise unavailable:", "_____no_output_____" ] ], [ [ "def tf_serialize_example(f0,f1,f2,f3):\n tf_string = tf.py_function(\n serialize_example,\n (f0,f1,f2,f3), # pass these args to the above function.\n tf.string) # the return type is `tf.string`.\n return tf.reshape(tf_string, ()) # The result is a scalar", "_____no_output_____" ], [ "tf_serialize_example(f0,f1,f2,f3)", "_____no_output_____" ] ], [ [ "Apply this function to each element in the dataset:", "_____no_output_____" ] ], [ [ "serialized_features_dataset = features_dataset.map(tf_serialize_example)\nserialized_features_dataset", "_____no_output_____" ], [ "def generator():\n for features in features_dataset:\n yield serialize_example(*features)", "_____no_output_____" ], [ "serialized_features_dataset = tf.data.Dataset.from_generator(\n generator, output_types=tf.string, output_shapes=())", "_____no_output_____" ], [ "serialized_features_dataset", "_____no_output_____" ] ], [ [ "And write them to a TFRecord file:", "_____no_output_____" ] ], [ [ "filename = 'test.tfrecord'\nwriter = tf.data.experimental.TFRecordWriter(filename)\nwriter.write(serialized_features_dataset)", "_____no_output_____" ] ], [ [ "### Reading a TFRecord file", "_____no_output_____" ], [ "You can also read the TFRecord file using the `tf.data.TFRecordDataset` class.\n\nMore information on consuming TFRecord files using `tf.data` can be found [here](https://www.tensorflow.org/guide/datasets#consuming_tfrecord_data).\n\nUsing `TFRecordDataset`s can be useful for standardizing input data and optimizing performance.", "_____no_output_____" ] ], [ [ "filenames = [filename]\nraw_dataset = tf.data.TFRecordDataset(filenames)\nraw_dataset", "_____no_output_____" ] ], [ [ "At this point the dataset contains serialized `tf.train.Example` messages. When iterated over it returns these as scalar string tensors.\n\nUse the `.take` method to only show the first 10 records.\n\nNote: iterating over a `tf.data.Dataset` only works with eager execution enabled.", "_____no_output_____" ] ], [ [ "for raw_record in raw_dataset.take(10):\n print(repr(raw_record))", "_____no_output_____" ] ], [ [ "These tensors can be parsed using the function below. Note that the `feature_description` is necessary here because datasets use graph-execution, and need this description to build their shape and type signature:", "_____no_output_____" ] ], [ [ "# Create a description of the features.\nfeature_description = {\n 'feature0': tf.io.FixedLenFeature([], tf.int64, default_value=0),\n 'feature1': tf.io.FixedLenFeature([], tf.int64, default_value=0),\n 'feature2': tf.io.FixedLenFeature([], tf.string, default_value=''),\n 'feature3': tf.io.FixedLenFeature([], tf.float32, default_value=0.0),\n}\n\ndef _parse_function(example_proto):\n # Parse the input `tf.Example` proto using the dictionary above.\n return tf.io.parse_single_example(example_proto, feature_description)", "_____no_output_____" ] ], [ [ "Alternatively, use `tf.parse example` to parse the whole batch at once. Apply this function to each item in the dataset using the `tf.data.Dataset.map` method:", "_____no_output_____" ] ], [ [ "parsed_dataset = raw_dataset.map(_parse_function)\nparsed_dataset", "_____no_output_____" ] ], [ [ "Use eager execution to display the observations in the dataset. There are 10,000 observations in this dataset, but you will only display the first 10. The data is displayed as a dictionary of features. Each item is a `tf.Tensor`, and the `numpy` element of this tensor displays the value of the feature:", "_____no_output_____" ] ], [ [ "for parsed_record in parsed_dataset.take(10):\n print(repr(parsed_record))", "_____no_output_____" ] ], [ [ "Here, the `tf.parse_example` function unpacks the `tf.Example` fields into standard tensors.", "_____no_output_____" ], [ "## TFRecord files in Python", "_____no_output_____" ], [ "The `tf.io` module also contains pure-Python functions for reading and writing TFRecord files.", "_____no_output_____" ], [ "### Writing a TFRecord file", "_____no_output_____" ], [ "Next, write the 10,000 observations to the file `test.tfrecord`. Each observation is converted to a `tf.Example` message, then written to file. You can then verify that the file `test.tfrecord` has been created:", "_____no_output_____" ] ], [ [ "# Write the `tf.Example` observations to the file.\nwith tf.io.TFRecordWriter(filename) as writer:\n for i in range(n_observations):\n example = serialize_example(feature0[i], feature1[i], feature2[i], feature3[i])\n writer.write(example)", "_____no_output_____" ], [ "!du -sh {filename}", "_____no_output_____" ] ], [ [ "### Reading a TFRecord file\n\nThese serialized tensors can be easily parsed using `tf.train.Example.ParseFromString`:", "_____no_output_____" ] ], [ [ "filenames = [filename]\nraw_dataset = tf.data.TFRecordDataset(filenames)\nraw_dataset", "_____no_output_____" ], [ "for raw_record in raw_dataset.take(1):\n example = tf.train.Example()\n example.ParseFromString(raw_record.numpy())\n print(example)", "_____no_output_____" ] ], [ [ "## Walkthrough: Reading and writing image data", "_____no_output_____" ], [ "This is an end-to-end example of how to read and write image data using TFRecords. Using an image as input data, you will write the data as a TFRecord file, then read the file back and display the image.\n\nThis can be useful if, for example, you want to use several models on the same input dataset. Instead of storing the image data raw, it can be preprocessed into the TFRecords format, and that can be used in all further processing and modelling.\n\nFirst, let's download [this image](https://commons.wikimedia.org/wiki/File:Felis_catus-cat_on_snow.jpg) of a cat in the snow and [this photo](https://upload.wikimedia.org/wikipedia/commons/f/fe/New_East_River_Bridge_from_Brooklyn_det.4a09796u.jpg) of the Williamsburg Bridge, NYC under construction.", "_____no_output_____" ], [ "### Fetch the images", "_____no_output_____" ] ], [ [ "cat_in_snow = tf.keras.utils.get_file('320px-Felis_catus-cat_on_snow.jpg', 'https://storage.googleapis.com/download.tensorflow.org/example_images/320px-Felis_catus-cat_on_snow.jpg')\nwilliamsburg_bridge = tf.keras.utils.get_file('194px-New_East_River_Bridge_from_Brooklyn_det.4a09796u.jpg','https://storage.googleapis.com/download.tensorflow.org/example_images/194px-New_East_River_Bridge_from_Brooklyn_det.4a09796u.jpg')", "_____no_output_____" ], [ "display.display(display.Image(filename=cat_in_snow))\ndisplay.display(display.HTML('Image cc-by: <a \"href=https://commons.wikimedia.org/wiki/File:Felis_catus-cat_on_snow.jpg\">Von.grzanka</a>'))", "_____no_output_____" ], [ "display.display(display.Image(filename=williamsburg_bridge))\ndisplay.display(display.HTML('<a \"href=https://commons.wikimedia.org/wiki/File:New_East_River_Bridge_from_Brooklyn_det.4a09796u.jpg\">From Wikimedia</a>'))", "_____no_output_____" ] ], [ [ "### Write the TFRecord file", "_____no_output_____" ], [ "As before, encode the features as types compatible with `tf.Example`. This stores the raw image string feature, as well as the height, width, depth, and arbitrary `label` feature. The latter is used when you write the file to distinguish between the cat image and the bridge image. Use `0` for the cat image, and `1` for the bridge image:", "_____no_output_____" ] ], [ [ "image_labels = {\n cat_in_snow : 0,\n williamsburg_bridge : 1,\n}", "_____no_output_____" ], [ "# This is an example, just using the cat image.\nimage_string = open(cat_in_snow, 'rb').read()\n\nlabel = image_labels[cat_in_snow]\n\n# Create a dictionary with features that may be relevant.\ndef image_example(image_string, label):\n image_shape = tf.image.decode_jpeg(image_string).shape\n\n feature = {\n 'height': _int64_feature(image_shape[0]),\n 'width': _int64_feature(image_shape[1]),\n 'depth': _int64_feature(image_shape[2]),\n 'label': _int64_feature(label),\n 'image_raw': _bytes_feature(image_string),\n }\n\n return tf.train.Example(features=tf.train.Features(feature=feature))\n\nfor line in str(image_example(image_string, label)).split('\\n')[:15]:\n print(line)\nprint('...')", "_____no_output_____" ] ], [ [ "Notice that all of the features are now stored in the `tf.Example` message. Next, functionalize the code above and write the example messages to a file named `images.tfrecords`:", "_____no_output_____" ] ], [ [ "# Write the raw image files to `images.tfrecords`.\n# First, process the two images into `tf.Example` messages.\n# Then, write to a `.tfrecords` file.\nrecord_file = 'images.tfrecords'\nwith tf.io.TFRecordWriter(record_file) as writer:\n for filename, label in image_labels.items():\n image_string = open(filename, 'rb').read()\n tf_example = image_example(image_string, label)\n writer.write(tf_example.SerializeToString())", "_____no_output_____" ], [ "!du -sh {record_file}", "_____no_output_____" ] ], [ [ "### Read the TFRecord file\n\nYou now have the file—`images.tfrecords`—and can now iterate over the records in it to read back what you wrote. Given that in this example you will only reproduce the image, the only feature you will need is the raw image string. Extract it using the getters described above, namely `example.features.feature['image_raw'].bytes_list.value[0]`. You can also use the labels to determine which record is the cat and which one is the bridge:", "_____no_output_____" ] ], [ [ "raw_image_dataset = tf.data.TFRecordDataset('images.tfrecords')\n\n# Create a dictionary describing the features.\nimage_feature_description = {\n 'height': tf.io.FixedLenFeature([], tf.int64),\n 'width': tf.io.FixedLenFeature([], tf.int64),\n 'depth': tf.io.FixedLenFeature([], tf.int64),\n 'label': tf.io.FixedLenFeature([], tf.int64),\n 'image_raw': tf.io.FixedLenFeature([], tf.string),\n}\n\ndef _parse_image_function(example_proto):\n # Parse the input tf.Example proto using the dictionary above.\n return tf.io.parse_single_example(example_proto, image_feature_description)\n\nparsed_image_dataset = raw_image_dataset.map(_parse_image_function)\nparsed_image_dataset", "_____no_output_____" ] ], [ [ "Recover the images from the TFRecord file:", "_____no_output_____" ] ], [ [ "for image_features in parsed_image_dataset:\n image_raw = image_features['image_raw'].numpy()\n display.display(display.Image(data=image_raw))", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# YAHOO電影爬蟲練習\n## 練習爬取電影放映資訊。必須逐步獲取電影的代號、放映地區、放映日期後，再送出查詢給伺服器。", "_____no_output_____" ] ], [ [ "import requests\nimport re\nfrom bs4 import BeautifulSoup", "_____no_output_____" ] ], [ [ "### 先搜尋全部的電影代號(ID)資訊", "_____no_output_____" ] ], [ [ "# 查看目前上映那些電影，並擷取出其ID資訊\nurl = 'https://movies.yahoo.com.tw/'\nresp = requests.get(url)\nresp.encoding = 'utf-8'\n# gggggg\nsoup = BeautifulSoup(resp.text, 'lxml')\nhtml = soup.find(\"select\", attrs={'name':'movie_id'})\nmovie_item = html.find_all(\"option\", attrs={'data-name':re.compile('.*')})\n\nfor p in movie_item:\n print(\"Movie: %s, ID: %s\" % (p[\"data-name\"], p[\"value\"]))", "Movie: 空中謎航, ID: 11152\nMovie: 致命天際線, ID: 11147\nMovie: 廢青四重奏, ID: 11130\nMovie: 妄想代理人：前篇, ID: 11102\nMovie: 水漾的女人, ID: 11065\nMovie: 午夜天鵝, ID: 11045\nMovie: 拆彈專家2, ID: 10986\nMovie: 杏林醫院, ID: 10781\nMovie: 靈魂急轉彎, ID: 11089\nMovie: 瑰麗卡萊爾：浮華紐約, ID: 11129\nMovie: 愛是您・愛是我, ID: 11123\nMovie: 來者弒客, ID: 11107\nMovie: 真愛鄰距離, ID: 11101\nMovie: 坑爹大作戰, ID: 11082\nMovie: 生為女人, ID: 10977\nMovie: 一家之煮, ID: 10955\nMovie: 腿, ID: 10934\nMovie: 鬼巢, ID: 11105\nMovie: 高校棋蹟, ID: 11099\nMovie: 戀愛進行弒, ID: 11080\nMovie: 85年的夏天, ID: 11076\nMovie: 順其自然的日子, ID: 11066\nMovie: 總是有個人在愛你, ID: 11063\nMovie: 神力女超人1984, ID: 10413\nMovie: 新解釋・三國志, ID: 11050\nMovie: 信用詐欺師JP：公主篇, ID: 11021\nMovie: 求婚好意外, ID: 10796\nMovie: 再見街貓BOB, ID: 11016\nMovie: 愛在午夜希臘時, ID: 11054\nMovie: 愛在黎明破曉時, ID: 11053\nMovie: 愛在日落巴黎時, ID: 11052\nMovie: 魔物獵人, ID: 10983\nMovie: 親愛的殺手, ID: 10861\nMovie: 未來的我們, ID: 11046\nMovie: 十二夜2：回到第零天, ID: 11035\nMovie: 尋找小魔女Doremi, ID: 11027\nMovie: 緝毒風暴, ID: 11023\nMovie: 古魯家族：新石代, ID: 10958\nMovie: 同學麥娜絲, ID: 10935\nMovie: 名偵探柯南：紅之校外旅行 鮮紅篇&戀紅篇, ID: 10887\nMovie: 我的媽媽開GAYBAR, ID: 10973\nMovie: 愛麗絲與夢幻島, ID: 11018\nMovie: 孤味, ID: 10477\nMovie: 聖荷西謀殺案, ID: 10990\nMovie: 入魔, ID: 10989\nMovie: 悄悄告訴她 經典數位修復, ID: 10911\nMovie: 鬼滅之刃劇場版 無限列車篇, ID: 10816\nMovie: 親愛的房客, ID: 10707\nMovie: 地下弒的秘密, ID: 10984\nMovie: 藥頭大媽, ID: 10951\nMovie: 看不見的目擊者, ID: 10946\nMovie: 無價之保, ID: 10959\nMovie: 倒數反擊, ID: 10906\nMovie: 阿公當家, ID: 10914\nMovie: 刻在你心底的名字, ID: 10902\nMovie: 急先鋒, ID: 10443\nMovie: 殺戮荒村, ID: 10903\nMovie: 消失的情人節, ID: 10870\nMovie: 中央車站：數位修復版, ID: 10907\nMovie: 海霧, ID: 10872\nMovie: 花木蘭, ID: 8632\nMovie: TENET天能, ID: 10433\nMovie: 聖雅各的天空, ID: 10877\nMovie: 看不見的證人, ID: 10873\nMovie: 可不可以，你也剛好喜歡我, ID: 10473\nMovie: 東京教父：4K數位修復版, ID: 10860\nMovie: 劇場版 新幹線變形機器人—來自未來的神速ALFA-X, ID: 10823\nMovie: 巴亞拉魔幻冒險, ID: 10851\nMovie: 怪胎, ID: 10733\nMovie: 魔王的女兒, ID: 10730\nMovie: 藍色恐懼：數位修復版, ID: 10775\nMovie: 角落小夥伴電影版：魔法繪本裡的新朋友, ID: 10647\nMovie: 一首搖滾上月球, ID: 4887\nMovie: 錢不夠用2, ID: 3026\n" ] ], [ [ "### 指定你有興趣的電影其ID，然後查詢其放映地區資訊。", "_____no_output_____" ] ], [ [ "# 參考前一個步驟中擷取到的ID資訊，並指定ID\nmovie_id = 10477", "_____no_output_____" ], [ "url = 'https://movies.yahoo.com.tw/api/v1/areas_by_movie_theater'\npayload = {'movie_id':str(movie_id)}\n\n# 模擬一個header\nheaders = {\n 'authority': 'movies.yahoo.com.tw',\n 'method': 'GET',\n 'path': '/api/v1/areas_by_movie_theater?movie_id=' + str(movie_id),\n 'scheme': 'https',\n 'accept': 'application/json, text/javascript, */*; q=0.01',\n 'accept-encoding': 'gzip, deflate, br',\n 'accept-language': 'zh-TW,zh;q=0.9,en-US;q=0.8,en;q=0.7,zh-CN;q=0.6',\n 'cookie': 'rxx=9s3x2fws06.1g16irnc&v=1; _ga=GA1.3.2056742944.1551651301; GUC=AQEBAQFczFpdm0IfmwSB&s=AQAAACoo4N5D&g=XMsVBw; BX=4hkdk1decm57t&b=3&s=mr; _ga=GA1.4.2056742944.1551651301; nexagesuid=82843256dd234e8e91aa73f2062f8218; browsed_movie=eyJpdiI6IlJXWWtiSWJaZlNGK2MxQnhscnVUYWc9PSIsInZhbHVlIjoiMXRhMmVHRXRIeUNjc1RBWDJzdGYwbnlIQURmWGsrcjJSMzhkbkcraDNJVUNIZEZsbzU3amlFcVZ1NzlmazJrTGpoMjVrbHk1YmpoRENXaHZTOUw1TmI2ZTZVWHdOejZQZm16RmVuMWlHTTJLaTZLVFZZVkFOMDlTd1wvSGltcytJIiwibWFjIjoiZWQ2ZjA4MmVjZmZlYjlmNjJmYmY2NGMyMDI0Njc0NWViYjVkOWE2NDg0N2RhODMxZjBjZDhiMmJhZTc2MDZhYiJ9; avi=eyJpdiI6Im1NeWFJRlVRWDR1endEcGRGUGJUbVE9PSIsInZhbHVlIjoickRpU3JuUytmcGl6cjF5OW0rNU9iZz09IiwibWFjIjoiY2VmY2NkNzZmM2NhNjY5YzlkOTcyNjE5OGEyMzU0NWYxOTdmMDRkMDY3OWNmMmZjOTMxYjc5MjI5N2Q5NGE5MiJ9; cmp=t=1559391030&j=0; _gid=GA1.4.779543841.1559391031; XSRF-TOKEN=eyJpdiI6IkhpS2hGcDRQaHlmWUJmaHdSS2Q2bHc9PSIsInZhbHVlIjoiOUVoNFk4OHI1UUZmUWRtYXhza0MyWjJSTlhlZ3RnT0VGeVJPN2JuczVRMGRFdWt2OUlsamVKeHRobFwvcHBGM0dhU3VyMXNGTHlsb2dVM2l0U1hpUGxBPT0iLCJtYWMiOiJkZWU4YzJhNjAxMTY3MzE4Y2ExNWIxYmE1ZjE1YWZlZTlhOTcyYjc4M2RlZGY4ZWNjZDYyMTA2NGYwZGViMzc2In0%3D; m_s=eyJpdiI6InpsZHZ2Tk1BZ0dxaHhETml1RjBnUXc9PSIsInZhbHVlIjoiSkNGeHUranRoXC85bDFiaDhySTJqNkJRcWdjWUxjeVRJSHVYZ1wvd2d4bWJZUTUrSHVDM0lUcW5KNHdETFZ4T1lieU81OUhzc1VoUXhZcWk0UDZSQXVFdz09IiwibWFjIjoiYmJkMDJkMDhlODIzMzcyMWY4M2NmYWNjNGVlOWRjMDIwZmVmNzAyMjE3Yzg3ZGY3ODBkZWEzZTI4MTI5ZWNmOSJ9; _gat=1; nexagesd=10',\n 'dnt': '1',\n 'mv-authorization': '21835b082e15b91a69b3851eec7b31b82ce82afb',\n 'referer': 'https://movies.yahoo.com.tw/',\n 'user-agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/74.0.3729.169 Safari/537.36',\n 'x-requested-with': 'XMLHttpRequest',\n}\n \nresp = requests.get(url, params=payload, headers=headers)\n#print(resp.json()) # 若有需要，列印出json原始碼\n\n# 這裡回傳的格式是JSON格式的資料，要解析JSON擷取資料\nfor p in resp.json():\n print('放映地區: {0}, 代號(area_id): {1}'.format(p['title'], p['area_id']))", "放映地區: 台北市, 代號(area_id): 28\n放映地區: 新北市, 代號(area_id): 8\n放映地區: 桃園, 代號(area_id): 16\n放映地區: 新竹, 代號(area_id): 20\n放映地區: 苗栗, 代號(area_id): 15\n放映地區: 台中, 代號(area_id): 2\n放映地區: 南投, 代號(area_id): 13\n放映地區: 嘉義, 代號(area_id): 21\n放映地區: 台南, 代號(area_id): 10\n放映地區: 高雄, 代號(area_id): 17\n放映地區: 宜蘭, 代號(area_id): 11\n放映地區: 花蓮, 代號(area_id): 12\n放映地區: 澎湖, 代號(area_id): 23\n" ] ], [ [ "### 指定你想要觀看的放映地區，查詢有上映電影的場次日期", "_____no_output_____" ] ], [ [ "# 指定放映地區\narea_id = 28", "_____no_output_____" ], [ "# 向網站發送請求\nurl = 'https://movies.yahoo.com.tw/movietime_result.html'\npayload = {'movie_id':str(movie_id), 'area_id':str(area_id)}\nresp = requests.get(url, params=payload)\nresp.encoding = 'utf-8'\n\nsoup = BeautifulSoup(resp.text, 'lxml')\nmovie_date = soup.find_all(\"label\", attrs={'for':re.compile(\"date_[\\d]\")})\n\n# 列印播放日期\nfor date in movie_date:\n print(\"%s %s\" % (date.p.string, date.h3.string))", "一月 1\n一月 2\n一月 3\n一月 4\n一月 5\n" ] ], [ [ "### 最後指定觀看的日期，查詢並列印出放映的電影院、放映類型(數位、3D、IMAX 3D...)、放映時間等資訊。", "_____no_output_____" ] ], [ [ "# 選定要觀看的日期\ndate = \"2019-08-21\"", "_____no_output_____" ], [ "# 向網站發送請求，獲取上映的電影院及時間資訊\nurl = \"https://movies.yahoo.com.tw/ajax/pc/get_schedule_by_movie\"\npayload = {'movie_id':str(movie_id),\n 'date':date,\n 'area_id':str(area_id),\n 'theater_id':'',\n 'datetime':'',\n 'movie_type_id':''}\n\nresp = requests.get(url, params=payload)\n#print(resp.json()['view']) # 若有需要，列印出json原始碼\n\nsoup = BeautifulSoup(resp.json()['view'], 'lxml')\nhtml = soup.find_all(\"ul\", attrs={'data-theater_name':re.compile(\".*\")})", "_____no_output_____" ], [ "'''\n\n 試著從上一步驟回傳的電影院資料中，擷取電影院名稱、影片放映類型以及時間表\n \n Your code here.\n\n'''\n", "----------------------------------------------------------------------\n電影院: 台北美麗華大直影城\n放映類型: 數位\n2019-08-21 09:00:00\n2019-08-21 11:10:00\n2019-08-21 13:15:00\n2019-08-21 15:20:00\n2019-08-21 19:30:00\n2019-08-21 21:40:00\n2019-08-21 22:30:00\n----------------------------------------------------------------------\n電影院: 台北新光影城\n放映類型: 數位\n2019-08-21 10:00:00\n2019-08-21 14:50:00\n2019-08-21 19:30:00\n----------------------------------------------------------------------\n電影院: 台北in89豪華數位影城\n放映類型: 數位\n2019-08-21 09:30:00\n2019-08-21 11:20:00\n2019-08-21 13:15:00\n2019-08-21 15:10:00\n2019-08-21 16:10:00\n2019-08-21 17:10:00\n2019-08-21 18:05:00\n2019-08-21 19:10:00\n2019-08-21 21:10:00\n2019-08-21 23:10:00\n2019-08-22 01:10:00\n----------------------------------------------------------------------\n電影院: 台北日新威秀影城\n放映類型: 數位\n2019-08-21 09:00:00\n2019-08-21 10:55:00\n2019-08-21 12:50:00\n2019-08-21 14:45:00\n2019-08-21 16:40:00\n2019-08-21 18:35:00\n2019-08-21 20:35:00\n----------------------------------------------------------------------\n電影院: 喜滿客絕色影城\n放映類型: 數位\n2019-08-21 10:00:00\n2019-08-21 11:55:00\n2019-08-21 13:50:00\n2019-08-21 15:45:00\n2019-08-21 17:40:00\n2019-08-21 19:35:00\n2019-08-21 21:30:00\n----------------------------------------------------------------------\n電影院: 台北信義威秀影城\n放映類型: 數位\n2019-08-21 09:00:00\n2019-08-21 11:00:00\n2019-08-21 13:00:00\n2019-08-21 15:00:00\n2019-08-21 17:00:00\n2019-08-21 19:00:00\n2019-08-21 21:00:00\n2019-08-21 23:00:00\n----------------------------------------------------------------------\n電影院: 喜滿客京華影城\n放映類型: 數位\n2019-08-21 10:30:00\n2019-08-21 12:30:00\n2019-08-21 14:30:00\n2019-08-21 16:30:00\n2019-08-21 18:30:00\n2019-08-21 20:30:00\n2019-08-21 22:30:00\n----------------------------------------------------------------------\n電影院: 京站威秀影城\n放映類型: 數位\n2019-08-21 09:00:00\n2019-08-21 11:00:00\n2019-08-21 13:00:00\n2019-08-21 15:00:00\n2019-08-21 17:00:00\n2019-08-21 19:00:00\n2019-08-21 21:00:00\n----------------------------------------------------------------------\n電影院: 喜樂時代影城南港店\n放映類型: 數位\n2019-08-21 10:20:00\n2019-08-21 11:10:00\n2019-08-21 12:20:00\n2019-08-21 13:10:00\n2019-08-21 14:20:00\n2019-08-21 15:10:00\n2019-08-21 16:20:00\n2019-08-21 17:10:00\n2019-08-21 18:20:00\n2019-08-21 19:15:00\n2019-08-21 20:20:00\n2019-08-21 21:15:00\n2019-08-21 22:20:00\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "empty" ] ] ]

[ "empty" ]

[ [ "empty" ] ]

[ "CC0-1.0", "BSD-3-Clause" ]

[ "CC0-1.0", "BSD-3-Clause" ]

[ "CC0-1.0", "BSD-3-Clause" ]

[ [ [ "# Flopy MODFLOW 6 (MF6) Support", "_____no_output_____" ], [ "The Flopy library contains classes for creating, saving, running, loading, and modifying MF6 simulations. The MF6 portion of the flopy library is located in:\n\n*flopy.mf6*\n\nWhile there are a number of classes in flopy.mf6, to get started you only need to use the main classes summarized below:\n\nflopy.mf6.MFSimulation \n* MODFLOW Simulation Class. Entry point into any MODFLOW simulation.\n\nflopy.mf6.ModflowGwf\n* MODFLOW Groundwater Flow Model Class. Represents a single model in a simulation.\n\nflopy.mf6.Modflow[pc]\n * MODFLOW package classes where [pc] is the abbreviation of the package name. Each package is a separate class. \n\nFor packages that are part of a groundwater flow model, the abbreviation begins with \"Gwf\". For example, \"flopy.mf6.ModflowGwfdis\" is the Discretization package.\n ", "_____no_output_____" ] ], [ [ "import os\nimport sys\nfrom shutil import copyfile\n\nimport numpy as np\nimport matplotlib as mpl\nimport matplotlib.pyplot as plt\n\ntry:\n import flopy\nexcept:\n fpth = os.path.abspath(os.path.join('..', '..'))\n sys.path.append(fpth)\n import flopy\n \nprint(sys.version)\nprint('numpy version: {}'.format(np.__version__))\nprint('matplotlib version: {}'.format(mpl.__version__))\nprint('flopy version: {}'.format(flopy.__version__))", "3.8.10 (default, May 19 2021, 11:01:55) \n[Clang 10.0.0 ]\nnumpy version: 1.19.2\nmatplotlib version: 3.4.2\nflopy version: 3.3.4\n" ] ], [ [ "# Creating a MF6 Simulation", "_____no_output_____" ], [ "A MF6 simulation is created by first creating a simulation object \"MFSimulation\". When you create the simulation object you can define the simulation's name, version, executable name, workspace path, and the name of the tdis file. All of these are optional parameters, and if not defined each one will default to the following:\n\nsim_name='modflowtest'\n\nversion='mf6'\n\nexe_name='mf6.exe'\n\nsim_ws='.'\n\nsim_tdis_file='modflow6.tdis'", "_____no_output_____" ] ], [ [ "import os\nimport sys\nfrom shutil import copyfile\ntry:\n import flopy\nexcept:\n fpth = os.path.abspath(os.path.join('..', '..'))\n sys.path.append(fpth)\n import flopy\n\nsim_name = 'example_sim'\nsim_path = os.path.join('data', 'example_project')\nsim = flopy.mf6.MFSimulation(sim_name=sim_name, version='mf6', exe_name='mf6', \n sim_ws=sim_path)", "_____no_output_____" ] ], [ [ "The next step is to create a tdis package object \"ModflowTdis\". The first parameter of the ModflowTdis class is a simulation object, which ties a ModflowTdis object to a specific simulation. The other parameters and their definitions can be found in the docstrings.", "_____no_output_____" ] ], [ [ "tdis = flopy.mf6.ModflowTdis(sim, pname='tdis', time_units='DAYS', nper=2, \n perioddata=[(1.0, 1, 1.0), (10.0, 5, 1.0)])", "_____no_output_____" ] ], [ [ "Next one or more models are created using the ModflowGwf class. The first parameter of the ModflowGwf class is the simulation object that the model will be a part of.", "_____no_output_____" ] ], [ [ "model_name = 'example_model'\nmodel = flopy.mf6.ModflowGwf(sim, modelname=model_name,\n model_nam_file='{}.nam'.format(model_name))", "_____no_output_____" ] ], [ [ "Next create one or more Iterative Model Solution (IMS) files.", "_____no_output_____" ] ], [ [ "ims_package = flopy.mf6.ModflowIms(sim, pname='ims', print_option='ALL',\n complexity='SIMPLE', outer_hclose=0.00001,\n outer_maximum=50, under_relaxation='NONE',\n inner_maximum=30, inner_hclose=0.00001,\n linear_acceleration='CG',\n preconditioner_levels=7,\n preconditioner_drop_tolerance=0.01,\n number_orthogonalizations=2)", "_____no_output_____" ] ], [ [ "Each ModflowGwf object needs to be associated with an ModflowIms object. This is done by calling the MFSimulation object's \"register_ims_package\" method. The first parameter in this method is the ModflowIms object and the second parameter is a list of model names (strings) for the models to be associated with the ModflowIms object.", "_____no_output_____" ] ], [ [ "sim.register_ims_package(ims_package, [model_name])", "_____no_output_____" ] ], [ [ "Next add packages to each model. The first package added needs to be a spatial discretization package since flopy uses information from the spatial discretization package to help you build other packages. There are three spatial discretization packages to choose from:\n\nDIS (ModflowGwfDis) - Structured discretization\nDISV (ModflowGwfdisv) - Discretization with vertices\nDISU (ModflowGwfdisu) - Unstructured discretization", "_____no_output_____" ] ], [ [ "dis_package = flopy.mf6.ModflowGwfdis(model, pname='dis', length_units='FEET', nlay=2,\n nrow=2, ncol=5, delr=500.0,\n delc=500.0,\n top=100.0, botm=[50.0, 20.0],\n filename='{}.dis'.format(model_name))", "_____no_output_____" ] ], [ [ "## Accessing Namefiles\n\nNamefiles are automatically built for you by flopy. However, there are some options contained in the namefiles that you may want to set. To get the namefile object access the name_file attribute in either a simulation or model object to get the simulation or model namefile.", "_____no_output_____" ] ], [ [ "# set the nocheck property in the simulation namefile\nsim.name_file.nocheck = True\n# set the print_input option in the model namefile\nmodel.name_file.print_input = True", "_____no_output_____" ] ], [ [ "## Specifying Options\n\nOption that appear alone are assigned a boolean value, like the print_input option above. Options that have additional optional parameters are assigned using a tuple, with the entries containing the names of the optional parameters to turn on. Use a tuple with an empty string to indicate no optional parameters and use a tuple with None to turn the option off. ", "_____no_output_____" ] ], [ [ "# Turn Newton option on with under relaxation \nmodel.name_file.newtonoptions = ('UNDER_RELAXATION')\n# Turn Newton option on without under relaxation\nmodel.name_file.newtonoptions = ('')\n# Turn off Newton option \nmodel.name_file.newtonoptions = (None)", "_____no_output_____" ] ], [ [ "## MFArray Templates\n\nLastly define all other packages needed. \n\nNote that flopy supports a number of ways to specify data for a package. A template, which defines the data array shape for you, can be used to specify the data. Templates are built by calling the empty of the data type you are building. For example, to build a template for k in the npf package you would call:\n\nModflowGwfnpf.k.empty()\n\nThe empty method for \"MFArray\" data templates (data templates whose size is based on the structure of the model grid) take up to four parameters:\n\n* model - The model object that the data is a part of. A valid model object with a discretization package is required in order to build the proper array dimensions. This parameter is required.\n\n* layered - True or false whether the data is layered or not.\n\n* data_storage_type_list - List of data storage types, one for each model layer. If the template is not layered, only one data storage type needs to be specified. There are three data storage types supported, internal_array, internal_constant, and external_file. \n\n* default_value - The initial value for the array.", "_____no_output_____" ] ], [ [ "# build a data template for k that stores the first layer as an internal array and the second\n# layer as a constant with the default value of k for all layers set to 100.0 \nlayer_storage_types = [flopy.mf6.data.mfdatastorage.DataStorageType.internal_array, \n flopy.mf6.data.mfdatastorage.DataStorageType.internal_constant]\nk_template = flopy.mf6.ModflowGwfnpf.k.empty(model, True, layer_storage_types, 100.0)\n# change the value of the second layer to 50.0\nk_template[0]['data'] = [65.0, 60.0, 55.0, 50.0, 45.0, 40.0, 35.0, 30.0, 25.0, 20.0]\nk_template[0]['factor'] = 1.5\nprint(k_template)\n# create npf package using the k template to define k\nnpf_package = flopy.mf6.ModflowGwfnpf(model, pname='npf', save_flows=True, icelltype=1, k=k_template)", "[{'factor': 1.5, 'iprn': 1, 'data': [65.0, 60.0, 55.0, 50.0, 45.0, 40.0, 35.0, 30.0, 25.0, 20.0]}, 100.0]\n" ] ], [ [ "## Specifying MFArray Data\n\nMFArray data can also be specified as a numpy array, a list of values, or a single value. Below strt (starting heads) are defined as a single value, 100.0, which is interpreted as an internal constant storage type of value 100.0. Strt could also be defined as a list defining a value for every model cell:\n\nstrt=[100.0, 100.0, 100.0, 100.0, 100.0, 100.0, 100.0, 100.0, 100.0, 100.0, \n 90.0, 90.0, 90.0, 90.0, 90.0, 90.0, 90.0, 90.0, 90.0, 90.0]\n \nOr as a list defining a value or values for each model layer:\n\nstrt=[100.0, 90.0]\n\nor:\n\nstrt=[[100.0], [90.0, 90.0, 90.0, 90.0, 90.0, 90.0, 90.0, 90.0, 90.0, 90.0]]\n\nMFArray data can also be stored in an external file by using a dictionary using the keys 'filename' to specify the file name relative to the model folder and 'data' to specific the data. The optional 'factor', 'iprn', and 'binary' keys may also be used.\n\nstrt={'filename': 'strt.txt', 'factor':1.0, 'data':[100.0, 100.0, 100.0, 100.0, 100.0, 100.0, 100.0, 100.0, 100.0, 100.0, \n 90.0, 90.0, 90.0, 90.0, 90.0, 90.0, 90.0, 90.0, 90.0, 90.0], 'binary': 'True'}\n \nIf the 'data' key is omitted from the dictionary flopy will try to read the data from an existing file 'filename'. Any relative paths for loading data from a file should specified relative to the MF6 simulation folder.", "_____no_output_____" ] ], [ [ "strt={'filename': 'strt.txt', 'factor':1.0, 'data':[100.0, 100.0, 100.0, 100.0, 100.0, 100.0, 100.0, 100.0, 100.0, 100.0, \n 90.0, 90.0, 90.0, 90.0, 90.0, 90.0, 90.0, 90.0, 90.0, 90.0], 'binary': 'True'}\nic_package = flopy.mf6.ModflowGwfic(model, pname='ic', strt=strt,\n filename='{}.ic'.format(model_name))\n# move external file data into model folder\nicv_data_path = os.path.join('..', 'data', 'mf6', 'notebooks', 'iconvert.txt')\ncopyfile(icv_data_path, os.path.join(sim_path, 'iconvert.txt'))\n# create storage package\nsto_package = flopy.mf6.ModflowGwfsto(model, pname='sto', save_flows=True, iconvert={'filename':'iconvert.txt'},\n ss=[0.000001, 0.000002], \n sy=[0.15, 0.14, 0.13, 0.12, 0.11, 0.11, 0.12, 0.13, 0.14, 0.15,\n 0.15, 0.14, 0.13, 0.12, 0.11, 0.11, 0.12, 0.13, 0.14, 0.15])", "_____no_output_____" ] ], [ [ "## MFList Templates\n\nFlopy supports specifying record and recarray \"MFList\" data in a number of ways. Templates can be created that define the shape of the data. The empty method for \"MFList\" data templates take up to 7 parameters.\n\n* model - The model object that the data is a part of. A valid model object with a discretization package is required in order to build the proper array dimensions. This parameter is required.\n\n* maxbound - The number of rows in the recarray. If not specified one row is returned.\n\n* aux_vars - List of auxiliary variable names. If not specified auxiliary variables are not used.\n\n* boundnames - True/False if boundnames is to be used.\n\n* nseg - Number of segments (only relevant for a few data types)\n\n* timeseries - True/False indicates that time series data will be used.\n\n* stress_periods - List of integer stress periods to be used (transient MFList data only). If not specified for transient data, template will only be defined for stress period 1. \n\nMFList transient data templates are numpy recarrays stored in a dictionary with the dictionary key an integer zero based stress period value (stress period - 1).\n\nIn the code below the well package is set up using a transient MFList template to help build the well's stress_periods. ", "_____no_output_____" ] ], [ [ "maxbound = 2\n# build a stress_period_data template with 2 wells over stress periods 1 and 2 with boundnames \n# and three aux variables\nwel_periodrec = flopy.mf6.ModflowGwfwel.stress_period_data.empty(model, maxbound=maxbound, boundnames=True, \n aux_vars=['var1', 'var2', 'var3'],\n stress_periods=[0,1])\n# define the two wells for stress period one\nwel_periodrec[0][0] = ((0,1,2), -50.0, -1, -2, -3, 'First Well')\nwel_periodrec[0][1] = ((1,1,4), -25.0, 2, 3, 4, 'Second Well')\n# define the two wells for stress period two\nwel_periodrec[1][0] = ((0,1,2), -200.0, -1, -2, -3, 'First Well')\nwel_periodrec[1][1] = ((1,1,4), -4000.0, 2, 3, 4, 'Second Well')\n# build the well package\nwel_package = flopy.mf6.ModflowGwfwel(model, pname='wel', print_input=True, print_flows=True,\n auxiliary=['var1', 'var2', 'var3'], maxbound=maxbound,\n stress_period_data=wel_periodrec, boundnames=True, save_flows=True)", "/Users/jdhughes/Documents/Development/flopy_git/flopy_fork/flopy/mf6/data/mfdatalist.py:1688: FutureWarning: elementwise == comparison failed and returning scalar instead; this will raise an error or perform elementwise comparison in the future.\n if \"check\" in list_item:\n" ] ], [ [ "## Cell IDs\n\nCell IDs always appear as tuples in an MFList. For a structured grid cell IDs appear as:\n\n(&lt;layer&gt;, &lt;row&gt;, &lt;column&gt;)\n\nFor vertice based grid cells IDs appear as:\n\n(&lt;layer&gt;, &lt;intralayer_cell_id&gt;)\n\nUnstructured grid cell IDs appear as:\n\n(&lt;cell_id&gt;)", "_____no_output_____" ], [ "## Specifying MFList Data\n\nMFList data can also be defined as a list of tuples, with each tuple being a row of the recarray. For transient data the list of tuples can be stored in a dictionary with the dictionary key an integer zero based stress period value. If only a list of tuples is specified for transient data, the data is assumed to apply to stress period 1. Additional stress periods can be added with the add_transient_key method. The code below defines saverecord and printrecord as a list of tuples.", "_____no_output_____" ] ], [ [ "# printrecord data as a list of tuples. since no stress\n# period is specified it will default to stress period 1\nprintrec_tuple_list = [('HEAD', 'ALL'), ('BUDGET', 'ALL')]\n# saverecord data as a dictionary of lists of tuples for \n# stress periods 1 and 2. \nsaverec_dict = {0:[('HEAD', 'ALL'), ('BUDGET', 'ALL')],1:[('HEAD', 'ALL'), ('BUDGET', 'ALL')]}\n# create oc package\noc_package = flopy.mf6.ModflowGwfoc(model, pname='oc', \n budget_filerecord=[('{}.cbc'.format(model_name),)],\n head_filerecord=[('{}.hds'.format(model_name),)],\n saverecord=saverec_dict,\n printrecord=printrec_tuple_list)\n# add stress period two to the print record\noc_package.printrecord.add_transient_key(1)\n# set the data for stress period two in the print record\noc_package.printrecord.set_data([('HEAD', 'ALL'), ('BUDGET', 'ALL')], 1)", "_____no_output_____" ] ], [ [ "### Specifying MFList Data in an External File \n\nMFList data can be specified in an external file using a dictionary with the 'filename' key. If the 'data' key is also included in the dictionary and is not None, flopy will create the file with the data contained in the 'data' key. The 'binary' key can be used to save data to a binary file ('binary': True). The code below creates a chd package which creates and references an external file containing data for stress period 1 and stores the data internally in the chd package file for stress period 2. ", "_____no_output_____" ] ], [ [ "stress_period_data = {0: {'filename': 'chd_sp1.dat', 'data': [[(0, 0, 0), 70.]]},\n 1: [[(0, 0, 0), 60.]]}\nchd = flopy.mf6.ModflowGwfchd(model, maxbound=1, stress_period_data=stress_period_data)", "_____no_output_____" ] ], [ [ "## Packages that Support both List-based and Array-based Data\n\nThe recharge and evapotranspiration packages can be specified using list-based or array-based input. The array packages have an \"a\" on the end of their name:\n\nModflowGwfrch - list based recharge package\nModflowGwfrcha - array based recharge package\nModflowGwfevt - list based evapotranspiration package\nModflowGwfevta - array based evapotranspiration package", "_____no_output_____" ] ], [ [ "rch_recarray = {0:[((0,0,0), 'rch_1'), ((1,1,1), 'rch_2')],\n 1:[((0,0,0), 'rch_1'), ((1,1,1), 'rch_2')]}\nrch_package = flopy.mf6.ModflowGwfrch(model, pname='rch', fixed_cell=True, print_input=True, \n maxbound=2, stress_period_data=rch_recarray)", "_____no_output_____" ] ], [ [ "## Utility Files (TS, TAS, OBS, TAB)\n\nUtility files, MF6 formatted files that reference by packages, include time series, time array series, observation, and tab files. The file names for utility files are specified using the package that references them. The utility files can be created in several ways. A simple case is demonstrated below. More detail is given in the flopy3_mf6_obs_ts_tas notebook. ", "_____no_output_____" ] ], [ [ "# build a time series array for the recharge package\nts_data = [(0.0, 0.015, 0.0017), (1.0, 0.016, 0.0019), (2.0, 0.012, 0.0015),\n (3.0, 0.020, 0.0014), (4.0, 0.015, 0.0021), (5.0, 0.013, 0.0012),\n (6.0, 0.022, 0.0012), (7.0, 0.016, 0.0014), (8.0, 0.013, 0.0011),\n (9.0, 0.021, 0.0011), (10.0, 0.017, 0.0016), (11.0, 0.012, 0.0015)]\nrch_package.ts.initialize(time_series_namerecord=['rch_1', 'rch_2'], \n timeseries=ts_data, filename='recharge_rates.ts',\n interpolation_methodrecord=['stepwise', 'stepwise'])\n\n# build an recharge observation package that outputs the western recharge to a binary file and the eastern\n# recharge to a text file\nobs_data = {('rch_west.csv', 'binary'): [('rch_1_1_1', 'RCH', (0, 0, 0)),\n ('rch_1_2_1', 'RCH', (0, 1, 0))],\n 'rch_east.csv': [('rch_1_1_5', 'RCH', (0, 0, 4)),\n ('rch_1_2_5', 'RCH', (0, 1, 4))]}\nrch_package.obs.initialize(filename='example_model.rch.obs', digits=10, \n print_input=True, continuous=obs_data)", "_____no_output_____" ] ], [ [ "# Saving and Running a MF6 Simulation", "_____no_output_____" ], [ "Saving and running a simulation are done with the MFSimulation class's write_simulation and run_simulation methods.", "_____no_output_____" ] ], [ [ "# write simulation to new location\nsim.write_simulation()\n\n# run simulation\nsim.run_simulation()", "writing simulation...\n writing simulation name file...\n writing simulation tdis package...\n writing ims package ims...\n writing model example_model...\n writing model name file...\n writing package dis...\n writing package npf...\n writing package ic...\n writing package sto...\n writing package wel...\n writing package oc...\n writing package chd_0...\n writing package rch...\n writing package ts_0...\n writing package obs_0...\nFloPy is using the following executable to run the model: /Users/jdhughes/.local/bin/mf6\n MODFLOW 6\n U.S. GEOLOGICAL SURVEY MODULAR HYDROLOGIC MODEL\n VERSION 6.2.2 07/30/2021\n\n MODFLOW 6 compiled Aug 01 2021 12:51:08 with IFORT compiler (ver. 19.10.3)\n\nThis software has been approved for release by the U.S. Geological \nSurvey (USGS). Although the software has been subjected to rigorous \nreview, the USGS reserves the right to update the software as needed \npursuant to further analysis and review. No warranty, expressed or \nimplied, is made by the USGS or the U.S. Government as to the \nfunctionality of the software and related material nor shall the \nfact of release constitute any such warranty. Furthermore, the \nsoftware is released on condition that neither the USGS nor the U.S. \nGovernment shall be held liable for any damages resulting from its \nauthorized or unauthorized use. Also refer to the USGS Water \nResources Software User Rights Notice for complete use, copyright, \nand distribution information.\n\n \n Run start date and time (yyyy/mm/dd hh:mm:ss): 2021/08/06 16:21:19\n \n Writing simulation list file: mfsim.lst\n Using Simulation name file: mfsim.nam\n \n" ] ], [ [ "# Exporting a MF6 Model", "_____no_output_____" ], [ "Exporting a MF6 model to a shapefile or netcdf is the same as exporting a MF2005 model.", "_____no_output_____" ] ], [ [ "# make directory\npth = os.path.join('data', 'netCDF_export')\nif not os.path.exists(pth):\n os.makedirs(pth)\n \n# export the dis package to a netcdf file\nmodel.dis.export(os.path.join(pth, 'dis.nc'))\n\n# export the botm array to a shapefile\nmodel.dis.botm.export(os.path.join(pth, 'botm.shp'))", "initialize_geometry::proj4_str = epsg:4326\n" ] ], [ [ "# Loading an Existing MF6 Simulation", "_____no_output_____" ], [ "Loading a simulation can be done with the flopy.mf6.MFSimulation.load static method.", "_____no_output_____" ] ], [ [ "# load the simulation\nloaded_sim = flopy.mf6.MFSimulation.load(sim_name, 'mf6', 'mf6', sim_path)", "loading simulation...\n loading simulation name file...\n loading tdis package...\n loading model gwf6...\n loading package dis...\n loading package npf...\n loading package ic...\n loading package sto...\n loading package wel...\n loading package oc...\n loading package chd...\n" ] ], [ [ "# Retrieving Data and Modifying an Existing MF6 Simulation", "_____no_output_____" ], [ "Data can be easily retrieved from a simulation. Data can be retrieved using two methods. One method is to retrieve the data object from a master simulation dictionary that keeps track of all the data. The master simulation dictionary is accessed by accessing a simulation's \"simulation_data\" property and then the \"mfdata\" property:\n\nsim.simulation_data.mfdata[<data path>]\n\nThe data path is the path to the data stored as a tuple containing the model name, package name, block name, and data name.\n\nThe second method is to get the data from the package object. If you do not already have the package object, you can work your way down the simulation structure, from the simulation to the correct model, to the correct package, and finally to the data object.\n\nThese methods are demonstrated in the code below. ", "_____no_output_____" ] ], [ [ "# get hydraulic conductivity data object from the data dictionary\nhk = sim.simulation_data.mfdata[(model_name, 'npf', 'griddata', 'k')]\n\n# get specific yield data object from the storage package\nsy = sto_package.sy\n\n# get the model object from the simulation object using the get_model method, \n# which takes a string with the model's name and returns the model object\nmdl = sim.get_model(model_name)\n# get the package object from the model mobject using the get_package method,\n# which takes a string with the package's name or type\nic = mdl.get_package('ic')\n# get the data object from the initial condition package object\nstrt = ic.strt", "_____no_output_____" ] ], [ [ "Once you have the appropriate data object there are a number methods to retrieve data from that object. Data retrieved can either be the data as it appears in the model file or the data with any factor specified in the model file applied to it. To get the raw data without applying a factor use the get_data method. To get the data with the factor already applied use .array.\n\nNote that MFArray data is always a copy of the data stored by flopy. Modifying the copy of the flopy data will have no affect on the data stored in flopy. Non-constant internal MFList data is returned as a reference to a numpy recarray. Modifying this recarray will modify the data stored in flopy. ", "_____no_output_____" ] ], [ [ "# get the data without applying any factor\nhk_data_no_factor = hk.get_data()\nprint('Data without factor:\\n{}\\n'.format(hk_data_no_factor))\n\n# get data with factor applied\nhk_data_factor = hk.array\nprint('Data with factor:\\n{}\\n'.format(hk_data_factor))", "Data without factor:\n[[[ 65. 60. 55. 50. 45.]\n [ 40. 35. 30. 25. 20.]]\n\n [[100. 100. 100. 100. 100.]\n [100. 100. 100. 100. 100.]]]\n\nData with factor:\n[[[ 97.5 90. 82.5 75. 67.5]\n [ 60. 52.5 45. 37.5 30. ]]\n\n [[100. 100. 100. 100. 100. ]\n [100. 100. 100. 100. 100. ]]]\n\n" ] ], [ [ "Data can also be retrieved from the data object using []. For unlayered data the [] can be used to slice the data.", "_____no_output_____" ] ], [ [ "# slice layer one row two\nprint('SY slice of layer on row two\\n{}\\n'.format(sy[0,:,2]))", "SY slice of layer on row two\n[0.13 0.13]\n\n" ] ], [ [ "For layered data specify the layer number within the brackets. This will return a \"LayerStorage\" object which let's you change attributes of an individual layer.", "_____no_output_____" ] ], [ [ "# get layer one LayerStorage object\nhk_layer_one = hk[0]\n# change the print code and factor for layer one\nhk_layer_one.iprn = '2'\nhk_layer_one.factor = 1.1\nprint('Layer one data without factor:\\n{}\\n'.format(hk_layer_one.get_data()))\nprint('Data with new factor:\\n{}\\n'.format(hk.array))", "Layer one data without factor:\n[[65. 60. 55. 50. 45.]\n [40. 35. 30. 25. 20.]]\n\nData with new factor:\n[[[ 71.5 66. 60.5 55. 49.5]\n [ 44. 38.5 33. 27.5 22. ]]\n\n [[100. 100. 100. 100. 100. ]\n [100. 100. 100. 100. 100. ]]]\n\n" ] ], [ [ "## Modifying Data\n\nData can be modified in several ways. One way is to set data for a given layer within a LayerStorage object, like the one accessed in the code above. Another way is to set the data attribute to the new data. Yet another way is to call the data object's set_data method.", "_____no_output_____" ] ], [ [ "# set data within a LayerStorage object\nhk_layer_one.set_data([120.0, 100.0, 80.0, 70.0, 60.0, 50.0, 40.0, 30.0, 25.0, 20.0])\nprint('New HK data no factor:\\n{}\\n'.format(hk.get_data()))\n# set data attribute to new data\nic_package.strt = 150.0\nprint('New strt values:\\n{}\\n'.format(ic_package.strt.array))\n# call set_data\nsto_package.ss.set_data([0.000003, 0.000004])\nprint('New ss values:\\n{}\\n'.format(sto_package.ss.array))", "New HK data no factor:\n[[[120. 100. 80. 70. 60.]\n [ 50. 40. 30. 25. 20.]]\n\n [[100. 100. 100. 100. 100.]\n [100. 100. 100. 100. 100.]]]\n\nNew strt values:\n[[[150. 150. 150. 150. 150.]\n [150. 150. 150. 150. 150.]]\n\n [[150. 150. 150. 150. 150.]\n [150. 150. 150. 150. 150.]]]\n\nNew ss values:\n[[[3.e-06 3.e-06 3.e-06 3.e-06 3.e-06]\n [3.e-06 3.e-06 3.e-06 3.e-06 3.e-06]]\n\n [[4.e-06 4.e-06 4.e-06 4.e-06 4.e-06]\n [4.e-06 4.e-06 4.e-06 4.e-06 4.e-06]]]\n\n" ] ], [ [ "## Modifying the Simulation Path\n\nThe simulation path folder can be changed by using the set_sim_path method in the MFFileMgmt object. The MFFileMgmt object can be obtained from the simulation object through properties:\n\nsim.simulation_data.mfpath", "_____no_output_____" ] ], [ [ "# create new path\nsave_folder = os.path.join(sim_path, 'sim_modified')\n# change simulation path\nsim.simulation_data.mfpath.set_sim_path(save_folder)\n# create folder\nif not os.path.isdir(save_folder):\n os.makedirs(save_folder)", "WARNING: MFFileMgt's set_sim_path has been deprecated. Please use MFSimulation's set_sim_path in the future.\n" ] ], [ [ "## Adding a Model Relative Path\n\nA model relative path lets you put all of the files associated with a model in a folder relative to the simulation folder. Warning, this will override all of your file paths to model package files and will also override any relative file paths to external model data files. ", "_____no_output_____" ] ], [ [ "# Change path of model files relative to the simulation folder\nmodel.set_model_relative_path('model_folder')\n\n# create folder\nif not os.path.isdir(save_folder):\n os.makedirs(os.path.join(save_folder,'model_folder'))\n\n# write simulation to new folder\nsim.write_simulation()\n\n# run simulation from new folder\nsim.run_simulation() ", "writing simulation...\n writing simulation name file...\n writing simulation tdis package...\n writing ims package ims...\n writing model example_model...\n writing model name file...\n writing package dis...\n writing package npf...\n writing package ic...\n writing package sto...\n writing package wel...\n writing package oc...\n writing package chd_0...\n writing package rch...\n writing package ts_0...\n writing package obs_0...\nFloPy is using the following executable to run the model: /Users/jdhughes/.local/bin/mf6\n" ] ], [ [ "## Post-Processing the Results\n\nResults can be retrieved from the master simulation dictionary. Results are retrieved from the master simulation dictionary with using a tuple key that identifies the data to be retrieved. For head data use the key\n\n('&lt;model name&gt;', 'HDS', 'HEAD')\n\nwhere &lt;model name&gt; is the name of your model. For cell by cell budget data use the key\n\n('&lt;model name&gt;', 'CBC', '&lt;flow data name&gt;')\n\nwhere &lt;flow data name&gt; is the name of the flow data to be retrieved (ex. 'FLOW-JA-FACE'). All available output keys can be retrieved using the output_keys method.", "_____no_output_____" ] ], [ [ "keys = sim.simulation_data.mfdata.output_keys()", "('example_model', 'CBC', 'STO-SS')\n('example_model', 'CBC', 'STO-SY')\n('example_model', 'CBC', 'FLOW-JA-FACE')\n('example_model', 'CBC', 'WEL')\n('example_model', 'HDS', 'HEAD')\n" ] ], [ [ "The entries in the list above are keys for data in the head file \"HDS\" and data in cell by cell flow file \"CBC\". Keys in this list are not guaranteed to be in any particular order. The code below uses the head file key to retrieve head data and then plots head data using matplotlib.", "_____no_output_____" ] ], [ [ "import matplotlib.pyplot as plt\nimport numpy as np\n\n# get all head data\nhead = sim.simulation_data.mfdata['example_model', 'HDS', 'HEAD']\n# get the head data from the end of the model run\nhead_end = head[-1]\n# plot the head data from the end of the model run\nlevels = np.arange(160,162,1)\nextent = (0.0, 1000.0, 2500.0, 0.0)\nplt.contour(head_end[0, :, :],extent=extent)\nplt.show()", "_____no_output_____" ] ], [ [ "Results can also be retrieved using the existing binaryfile method.", "_____no_output_____" ] ], [ [ "# get head data using old flopy method\nhds_path = os.path.join(sim_path, model_name + '.hds')\nhds = flopy.utils.HeadFile(hds_path)\n# get heads after 1.0 days\nhead = hds.get_data(totim=1.0)\n# plot head data\nplt.contour(head[0, :, :],extent=extent)\nplt.show()", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Regression\nRegression is fundamentally a way to estimate an independent variable based on its relationships to predictor variables. This can be done both linearly and non-linearly with a single to many predictor variables. However, there are certain assumptions that must be satisfied in order for these results to be trustworthy.", "_____no_output_____" ] ], [ [ "#Import Packages\nimport numpy as np\nimport pandas as pd\n\n#Plotting\nimport seaborn as sns\nsns.set(rc={'figure.figsize':(11,8)})\nfrom matplotlib import pyplot as plt\nfrom mpl_toolkits.mplot3d import Axes3D\nimport matplotlib.ticker as ticker\n\n#Stats\nfrom scipy import stats\nimport statsmodels.api as sm\nimport statsmodels.formula.api as smf\nfrom statsmodels.stats.outliers_influence import variance_inflation_factor\nfrom statsmodels.tools.tools import add_constant\nfrom sklearn.linear_model import LinearRegression\nfrom sklearn.datasets import make_regression\nimport statistics", "_____no_output_____" ] ], [ [ "## Linear Regression\n\nRecall $y=mx+b$, in this case we can see that $y$ is defined by a pair of constants and $x$. Similarly in regression, we have our independent target variable being estimated by a set of derived constants and the input variable(s). Since what comes out of the model is not necessarily what was observed, we call the prediction of the model $\\hat{y}$. The equation is then: $\\hat{y}=\\hat{\\beta_0} + \\hat{\\beta_1}x_1 + ... + \\hat{\\beta_n}x_n + \\epsilon$ for n predictors, where the $\\hat{\\beta_i}$ are the estimated coefficients and constant.\n\nIn the case of linear regression, the line will have the equation:$\\hat{y}=\\hat{\\beta_0}+\\hat{\\beta_1}x_1+\\epsilon$, with $\\epsilon$ being the distance between the observation and the prediction, i.e. $y-\\hat{y}$.\n\nNow the question is \"how do we determine what the $\\hat{\\beta_0}\\ ...\\ \\hat{\\beta_n}$ should be?\" or rather \"how can we manipulate the $\\hat{\\beta_0}\\ ...\\ \\hat{\\beta_n}$ to minimize error?\"\n\n$\\epsilon$ is defined as the difference between reality and prediction, so minimizing this distance by itself would be a good starting place. However, since we generally want to increase punishment to the algorithm for missing points by a large margin, $punishment = (y-\\hat{y})^2 = \\epsilon^2$ is used. More formally, for a series of $m$ data points, the Sum of Squared Errors is defined as follows: $SSE=\\sum_{i=1}^{m}(y_i-\\hat{y_i})^2$ for $m$ predictions. This makes the Mean Squared Error $\\frac{SSE}{m}=MSE$. So, if we minimize mean squared error ($MSE$) we find the optimal line.", "_____no_output_____" ] ], [ [ "#Create a function to predict our line\ndef predict(x,slope,intercept):\n return (slope*x+intercept)", "_____no_output_____" ], [ "#Generate Data\nX = np.random.uniform(10,size=10)\nY = list(map(lambda x: x+ 2*np.random.randn(1)[0], X)) #f(x) = x + 2N\n\n#Graph Cloud\nsns.relplot(x='X', y='Y', data=pd.DataFrame({'X':X,'Y':Y}),color='darkblue', zorder=10)", "_____no_output_____" ] ], [ [ "Now, with a regression line", "_____no_output_____" ] ], [ [ "#Get Regression Results\nslope, intercept, r_value, p_value, std_err = stats.linregress(X,Y)\n\n#Create Regression Line\nlx = [min(X)-1,max(X)+1] #X-Bounds\nly = [predict(lx[0],slope,intercept),predict(lx[1],slope,intercept)] #Predictions at Bounds\n\n#Graph\nsns.relplot(x='X', y='Y', data=pd.DataFrame({'X':X,'Y':Y}),color='darkblue', zorder=10)\nsns.lineplot(lx, ly, color='r', zorder=5)", "_____no_output_____" ] ], [ [ "Then, find the $y_i-\\hat{y_i}$", "_____no_output_____" ] ], [ [ "#Plot Background\nsns.relplot(x='X', y='Y', data=pd.DataFrame({'X':X,'Y':Y}),color='darkblue', zorder=10)\nsns.lineplot(lx, ly, color='r', zorder=5)\n\n#Plot Distances\nfor i in range(len(X)):\n plt.plot([X[i],X[i]],[Y[i],predict(X[i],slope,intercept)],color = 'royalblue',linestyle=':',zorder=0)", "_____no_output_____" ] ], [ [ "Finally, calculate the MSE", "_____no_output_____" ] ], [ [ "#Setup\nfig, axs = plt.subplots(ncols=3, figsize=(15,5))\n\n#Calculations\nxy = pd.DataFrame({'X':X,'Y':Y}).sort_values('Y')\nxy['Y`'] = list(map(lambda x,y: ((predict(x,slope,intercept))), xy['X'], xy['Y']))\nxy['E'] = abs(xy['Y'] - xy['Y`'])\nxy = xy.sort_values(by=['E'])\n\n#Plot First Graph\nfor i in range(len(xy)):\n axs[0].plot([i,i],[0,xy.iloc[i,3]], color='royalblue',linestyle='-')\naxs[0].set_title('Sorted Errors')\n\n#Plot Second Graph\nfor i in range(len(xy)):\n axs[1].plot([i,i],[0,xy.iloc[i,3]**2], color='royalblue',linestyle='-')\naxs[1].set_title('Sorted Squared Errors')\n\n#Plot Third Graph\ntotal = min(xy['E'])\nfor i in range(len(xy)):\n deltah = xy.iloc[i,3]**2\n axs[2].plot([i,i],[0,xy.iloc[i,3]**2], color='royalblue',linestyle='-')\n axs[2].plot([max(0,i-1),i],[total,total + deltah], color='red',linestyle='-',marker='o')\n total+=deltah\naxs[2].set_title('Running Sum of Squared Erros')\nplt.show()\n\n#Calculate MSE\nMSE = statistics.mean(list(map(lambda e: e**2, xy['E'])))\nprint('Sum of Squared Errors:',round(total-min(xy['E']),3))\nprint('Total Observations:',len(xy))\nprint('Mean Squared Error:',round(MSE,4))", "_____no_output_____" ] ], [ [ "Now that we have a measure of success, we would like to develop some intuition and a function for how good we are predicting relative to perfect. If we were to imagine a straight line exactly through the mean of the Y-value (target value) of the cloud, we can imagine that half the points are above and half below. This calculation would give an $SSE$ of $\\sum_{i=1}^{n}(y_i-\\bar{y_i})^2$, or how this will be defined hereon the $SST$ or Sum of Squared Total Error. This guess is assumed to be the baseline guess you can make, and anything better is considered to be an improvement. As such, the ratio of $\\frac{SSE}{SST}$ models success relative to a benchmark. Functioning under the assumption that $0\\leq SSE\\leq SST$, $1-\\frac{SSE}{SST}$ gives a [0,1] interval of success generally known as $R^2$.\n\nLet's calculate.", "_____no_output_____" ] ], [ [ "#Create Data, Figures\nX = np.random.uniform(10,size=10)\nY = list(map(lambda x: x+ len(X)/10*np.random.randn(1)[0], X))\navgy = statistics.mean(Y)\nfig, axs = plt.subplots(ncols=2,figsize=(10,5))\n\n#Calculate Regressions\nslope, intercept, r_value, p_value, std_err = stats.linregress(X,Y)\nlx = [min(X)-1,max(X)+1]\nly = [predict(lx[0],slope,intercept),predict(lx[1],slope,intercept)]\navgy = statistics.mean(Y)\n\n#Calculate and Format MSE\nMSEr = 'MSE: '+str(round(statistics.mean(list(map(lambda x, y: (y-(predict(x,slope,intercept)))**2, X, Y))),3))\nMSEa = 'MSE: '+str(round(statistics.mean(list(map(lambda y: (y-avgy)**2, Y))),3))\n\n#Create Scatter And Lines\naxs[0].scatter(X,Y, color='darkblue', zorder=10)\naxs[1].scatter(X,Y, color='darkblue', zorder=10)\naxs[0].plot(lx, ly, color='r', zorder=5, label=MSEr)\naxs[1].plot(lx, [avgy,avgy], color='lightslategray', label=MSEa)\n\n#Create Dotted Lines\nfor i in range(len(X)):\n axs[0].plot([X[i],X[i]],[Y[i],predict(X[i],slope,intercept)],color = 'red',linestyle=':',zorder=0)\n axs[1].plot([X[i],X[i]],[Y[i],avgy],color = 'dimgray',linestyle=':',zorder=0) \n\n#Calculate R2\nR2r = 'Linear Regression: R-squared = '+str(round(r_value,3))\nSSTa = sum(list(map(lambda y: (y-statistics.mean(Y))**2, Y)))\nSSEa = sum(list(map(lambda y: (y-statistics.mean(Y))**2, Y)))\nR2a = 'Linear Regression: R-squared = '+str(round(1 - SSEa/SSTa,3))\n\n#Format\naxs[0].set(xlabel='X',ylabel='Y',title=R2r)\naxs[1].set(xlabel='X',ylabel='Y',title=R2a)\naxs[0].legend()\naxs[1].legend()\n\n#Paint Graphs\nplt.show()", "_____no_output_____" ] ], [ [ "Thankfully, we don't have to do this manually. To repeat this process and get out all the same information automatically, we can use SciPy, a scientific modeling library for python. In this case, as you saw above, we are using the linregress function to solve the linear system of equations that gives us the minimal MSE. Let's see how this would be implemented in practice.", "_____no_output_____" ] ], [ [ "X = np.random.randn(150).tolist()\nY = list(map(lambda x: x + np.random.randn(1)[0]/1.5, X)) #f(x) = x + N(0,2/3)\n\nreg = stats.linregress(X,Y)\ntitle = 'Linear Regression R-Squared: '+str(round(reg.rvalue,3))\n\nfig = sns.lmplot(x='X',y='Y',data=pd.DataFrame({'X':X,'Y':Y}),ci=False).set(title=title)\nfig.ax.get_lines()[0].set_color('red')", "_____no_output_____" ] ], [ [ "However, not all data can be regressed. There are four criterian that must be met in order for the assumptions necessary for regression to be satisfied. These can be generally analyzed through residuals, or $y-\\hat{y}$.\n\n1) Linearity\n\n2) Independence\n\n3) Homoscedasticity\n\n4) Normality\n\nWe will go through each one a describe how to ensure your data is acceptable.", "_____no_output_____" ], [ "### Linearity\nThis is essentially that the relationship between the variables in linear in nature\n\n#### Method 1: Residuals vs. Observed, what you are looking for is any pattern to the data.", "_____no_output_____" ] ], [ [ "#Data\nX = np.random.randn(300).tolist()", "_____no_output_____" ], [ "#Setup\nfig, axs = plt.subplots(ncols=2,figsize=(10,5))\n\n#Accepted\nY = list(map(lambda x: x + np.random.randn(1)[0]/1.5, X))\nsns.residplot(x='X',y='Y',data=pd.DataFrame({'X':X,'Y':Y})\n ,ax=axs[0]).set_title('Residuals of Linear Regression: Accepted')\n\n#Rejected\nY = list(map(lambda x: x**2 + np.random.randn(1)[0]/8, X))\nsns.residplot(x='X',y='Y',data=pd.DataFrame({'X':X,'Y':Y})\n ,ax=axs[1], color='b').set_title('Residuals of Linear Regression: Rejected')", "_____no_output_____" ] ], [ [ "#### Method 2: Observed vs. Predicted, looking for patterns", "_____no_output_____" ] ], [ [ "#Setup\nfig, axs = plt.subplots(ncols=2,figsize=(10,5))\n\n#Accept\nY = list(map(lambda x: x + np.random.randn(1)[0]/1.5, X))\nslope, intercept, r_value, p_value, std_err = stats.linregress(X,Y)\nYp = list(map(lambda x: predict(x,slope,intercept), X))\n\nsns.scatterplot(x='Y',y='Yp',data=pd.DataFrame({'Y':Y,'Yp':Yp}), ax=axs[0])\naxs[0].set(title='Predictions versus Observations: Accepted')\n\n#Reject\nY = list(map(lambda x: x**2 + np.random.randn(1)[0]/1.5, X))\nslope, intercept, r_value, p_value, std_err = stats.linregress(X,Y)\nYp = list(map(lambda x: predict(x,slope,intercept), X))\n\nsns.scatterplot(x='Y',y='Yp',data=pd.DataFrame({'Y':Y,'Yp':Yp}), ax=axs[1])\naxs[1].set(title='Predictions versus Observations: Rejected')", "_____no_output_____" ] ], [ [ "### Independence\nThis looks at whether the $\\epsilon$ is correlated to the predictors either through sorting or other methods. This is a serious consideration for time series data, but we will focus on non-time series for now. \n\n#### Method 1) Same as above. Compare the order of observations agains the residuals, there should be no pattern\n\n#### Method 2) For more than 1 predictor, use VIF scores to check for multicolinearity amongst the variables\nThere is no hard and fast rule for kicking variables out due to VIF scores, but a good rule of thumb is anything more than 10 needs should likely need to be dealt with and anything greater than 5 should at least be looked at.", "_____no_output_____" ] ], [ [ "#Calculate Scores\nA = np.random.randn(150).tolist()\nB = list(map(lambda a: a + np.random.randn(1)[0]/20, A))\nC = np.random.uniform(1,10,size=150).tolist()\n\ndata = pd.DataFrame({'A':A,'B':B,'C':C})\ndata = add_constant(data)\n\nVIFs = pd.DataFrame([variance_inflation_factor(data.values, i) for i in range(data.shape[1])], index=data.columns,\n columns=['VIF Score'])\nprint('Accept if less than 5, Reject if more than 10\\n')\nprint(VIFs)", "Accept if less than 5, Reject if more than 10\n\n VIF Score\nconst 5.102457\nA 431.508540\nB 431.471557\nC 1.005629\n" ] ], [ [ "### Homoscedasticity\nCheck for if the magnitude of the residuals is the same over time or observations\n\n#### Method 1) Ordered Plot", "_____no_output_____" ] ], [ [ "#Setup\nfig, axs = plt.subplots(ncols=2,figsize=(10,5))\n\n#Accept\nX = np.random.randn(150)\nY = np.random.randn(150)\n\nslope, intercept, r_value, p_value, std_err = stats.linregress(X, Y)\npredicted = X*slope+intercept\nresiduals = np.subtract(Y,predicted)\n\nsns.scatterplot(x='Predicted',y='Residuals', ax = axs[0],\n data=pd.DataFrame({'Predicted':predicted,'Residuals':residuals})).set_title('Accept')\n\n#Reject\nX = np.random.randn(150)\nY = list(map(lambda x: x**2 + np.random.randn(1)[0], X))\n\nslope, intercept, r_value, p_value, std_err = stats.linregress(X, Y)\npredicted = X*slope+intercept\nresiduals = np.subtract(Y,predicted)\n\nsns.scatterplot(x='Predicted',y='Residuals', ax = axs[1],\n data=pd.DataFrame({'Predicted':predicted,'Residuals':residuals})).set_title(\"Reject\")", "_____no_output_____" ] ], [ [ "### Normality\n\n#### Method 1) Q-Q Plot", "_____no_output_____" ] ], [ [ "fig, axs = plt.subplots(ncols=2,figsize=(10,5))\n\n#Accept\nX = np.random.randn(150)\nY = np.random.randn(150)\n\nslope, intercept, r_value, p_value, std_err = stats.linregress(X, Y)\npredicted = X*slope+intercept\nresiduals = np.subtract(Y,predicted)\n\nsm.qqplot(residuals, line='45',ax=axs[0])\n\n#Reject\nX = np.random.randn(150)\nY = list(map(lambda x: x**2 + np.random.randn(1)[0], X))\n\nslope, intercept, r_value, p_value, std_err = stats.linregress(X, Y)\npredicted = X*slope+intercept\nresiduals = np.subtract(Y,predicted)\n\nsm.qqplot(residuals, line='45',ax=axs[1])\n\nplt.show()", "_____no_output_____" ] ], [ [ "## Multiple Regression\n\nThis is exactly the same as before, but we can increase the number of predictors. For example, this is what it looks like to perform linear regression in free space, fitting a plane to a cloud of points.", "_____no_output_____" ] ], [ [ "#Generate Data\nX = np.random.randn(150)\nY = np.random.randn(150)\nZ = list(map(lambda y, x: 4*(y + x) + np.random.randn(1)[0], Y, X))\n\n#Regress\nreg = smf.ols(formula='Z ~ X+Y', data=pd.DataFrame({'X':X,'Y':Y,'Z':Z})).fit()\n\n#Calculate Plane\nxx = np.linspace(min(X), max(X), 8)\nyy = np.linspace(min(Y), max(Y), 8)\nXX, YY = np.meshgrid(xx, yy)\nZZ = XX*reg.params[1]+YY*reg.params[2]+reg.params[0]\n\n#Create the figure\nfig = plt.figure(figsize=(15,15))\nax = fig.add_subplot(111,projection='3d')\nax.set(xlabel='X',ylabel='Y',zlabel='Z')\nax.view_init(30, 315)\nax.set(title='Three Dimensional Visualization')\n\n#Plot Plane and Points\nax.scatter(X, Y, Z, color='blue')\nax.plot_wireframe(X=XX,Y=YY,Z=ZZ)", "_____no_output_____" ] ], [ [ "We are also still able to run all of the same tests as before, looking at the diagnostic plots.", "_____no_output_____" ] ], [ [ "#1) Linearity: Residuals vs. Observed\nresiduals = list(map(lambda x, y, z: z-(x*reg.params[1]+y*reg.params[2]+reg.params[0]), X,Y,Z))\nsns.relplot(x='Observed',y='Residuals',\n data=pd.DataFrame({'Observed':Z,'Residuals':residuals}))", "_____no_output_____" ], [ "#2) Independence: VIF Scores\ndata = pd.DataFrame({'X':X,'Y':Y})\ndata = add_constant(data)\nVIFs = pd.DataFrame([variance_inflation_factor(data.values, i) for i in range(data.shape[1])], index=data.columns,\n columns=['VIF Score'])\nprint('Accept if less than 5, Reject if more than 10\\n')\nprint(VIFs)", "Accept if less than 5, Reject if more than 10\n\n VIF Score\nconst 1.021775\nX 1.012041\nY 1.012041\n" ], [ "#3) Predicted vs. Residuals\npredicted = X*reg.params[1]+Y*reg.params[2]+reg.params[0]\nresiduals = list(map(lambda x, y, z: z-(x*reg.params[1]+y*reg.params[2]+reg.params[0]), X,Y,Z))\n\nsns.relplot(x='Predicted',y='Residuals',\n data=pd.DataFrame({'Predicted':predicted,'Residuals':residuals}))", "_____no_output_____" ], [ "#4) Normality: QQ Plot\nresiduals = np.array(list(map(lambda x, y, z: z-(x*reg.params[1]+y*reg.params[2]+reg.params[0]), X,Y,Z)))\nsm.qqplot(residuals, line='45')\nplt.show()", "_____no_output_____" ] ], [ [ "This can be done in infinitely many dimensions, but because we cannot visualize above three dimensions (at least statically), you can only investigate this with the diagnostic plots.", "_____no_output_____" ], [ "## Predictor Significance Analysis\nRemembering back to how you can statistically say whether two random variables are equivalent with a T test, we would like to know if the $\\hat{\\beta_i}$ is statistically different from zero. If it is, at whatever confidence level we choose, we can say that it is useful in predicting the target variable. This is generally done automatically by any regression engine and is stored in the form of a p-value. What the p-value says is the 1 - the probability that what was observed is different than what it is being tested against, in this case zero. In other words, the lower your p-value the higher the probability that the two values are different. When you hear the term statistical significance, what you are hearing is \"does the p-value cross our confidence boundary?\" This is often thought of as 0.1, 0.05, or 0.01 depending on the application. Each predictor will have it's own p-value, which can say whether or not it is useful in predicting the target variable. Let's see.", "_____no_output_____" ] ], [ [ "#Generate Data\nX = np.random.randn(150)\nY = np.random.randn(150)\nZ = list(map(lambda y, x: 4*(y + x) + np.random.randn(1)[0], Y, X))\n\n#Regress\nreg = smf.ols(formula='Z ~ X+Y', data=pd.DataFrame({'X':X,'Y':Y,'Z':Z})).fit()\n\n#Grab P-values and Coefficients\nconfint = pd.DataFrame(reg.conf_int())\npvalues = pd.DataFrame({'P-Value':reg.pvalues.round(7)})\nprint(pvalues)\nprint(confint)", " P-Value\nIntercept 0.240027\nX 0.000000\nY 0.000000\n 0 1\nIntercept -0.064758 0.256575\nX 3.851395 4.147178\nY 3.821662 4.162070\n" ] ], [ [ "This is generally much easier to do in purely statistical engines like R. We will cover that soon.", "_____no_output_____" ], [ "## Solving Regression\nThere are many ways that regression equations can be solved, one of these ways is through gradient descent, essentially adjusting a combination of variables until you find the minimum MSE. Let's visualize this. Note, this will take a minute or so to run as it calculating the MSE of a regression line for all the possibilities. What will develop is a smooth visualization of the MSEs. If you imagine placing a marble on the graph, where it roles is the direction to the best fit line. It will roll, and reverse, and roll, and reverse until it finally settles at the lowest point. This is gradient descent.", "_____no_output_____" ] ], [ [ "def getMSE(slope, intercept, X, Y):\n return statistics.mean(list(map(lambda x, y: (y-(predict(x,slope,intercept)))**2, X, Y)))\n\ndef getHMData(slope,intercept,X,Y,step=0.1,):\n slopeDist = slope[1]-slope[0]\n slopeSteps = int(slopeDist/step)\n \n interceptDist = intercept[1]-intercept[0]\n interceptSteps = int(interceptDist/step)\n \n data = []\n \n for i in range(slopeSteps):\n row = []\n for j in range(interceptSteps):\n row.append(getMSE(slope=slope[0]+step*i,intercept=intercept[0]+step*j,X=X,Y=Y))\n data.append(row)\n \n return pd.DataFrame(data)", "_____no_output_____" ], [ "#Set Limits\nincrement = 0.1\nslopes = [-10,10]\nnumslopes = (slopes[1]-slopes[0])/increment\nintercepts = [-10,10]\nnuminter = (intercepts[1]-intercepts[0])/increment\n\n#Get Data\nX = np.random.randn(500)-2*np.random.randn(1)[0]\nY = list(map(lambda x: np.random.uniform(1,4)*x + np.random.randn(1)[0], X))\ndata = getHMData(slope=slopes,intercept=intercepts,X=X,Y=Y,step=increment)\n\n#Format Labels\ndata = data.set_index(np.linspace(slopes[0],slopes[1],int(numslopes)).round(1))\ndata.columns = np.linspace(intercepts[0],intercepts[1],int(numinter)).round(1)\n\n#Heat Map of MSE\nfig, axs = plt.subplots(ncols=2,figsize=(20,10))\nsns.heatmap(data.iloc[::-1,:],cmap=\"RdYlGn_r\",ax = axs[1]).set(title='MSE over Slopes and Intercepts')\naxs[1].set(xlabel='Intercept',ylabel='Slope')\n\n#Regression Plot\nsns.regplot(x='X',y='Y',data=pd.DataFrame({'X':X,'Y':Y}), ax = axs[0])\nslope, intercept, r_value, p_value, std_err = stats.linregress(X, Y)\ntitle = ('Regression\\nSlope: '+str(slope.round(3))+', Intercept: '+\n str(intercept.round(3))+', R-Squared: '+str(r_value.round(3)))\naxs[0].set_title(title)\n\n#Plot Over Heatmap\naxs[1].hlines([slope*-10+100], *axs[1].get_xlim(),linestyles='dashed',colors='indigo')\naxs[1].vlines([intercept*10+100], *axs[1].get_ylim(),linestyles='dashed',colors='indigo')\naxs[1].scatter(intercept*10+100,slope*-10+100,zorder=10,c='indigo',s=50)", "_____no_output_____" ] ], [ [ "## Your turn to solve a problem.\n\nThe below code will generate for you 3 sets of values: 'X1','X2' - Predictors, 'Y' - Target\nYou will need to generate a regression. Then, you should check the predictors to see which one is significant. After this, you will need to re-run the regression and confirm if the data satisfies our regression assumption. Finally, you should create a seaborn plot of your regression.", "_____no_output_____" ] ], [ [ "Xs, Y = make_regression(n_samples=300, n_features=2, n_informative=1, noise=1)\ndata = pd.DataFrame(np.column_stack([Xs,Y]), columns=['X1','X2','Y'])", "_____no_output_____" ], [ "reg = smf.ols(formula='Y~X1+X2', data=pd.DataFrame({'X1':data['X1'],'X2':data['X2'],'Y':data['Y']})).fit()", "_____no_output_____" ], [ "confint = pd.DataFrame(reg.conf_int())\npvalues = pd.DataFrame({'P-Value':reg.pvalues.round(7)})\nprint(confint)\nprint(pvalues)", " 0 1\nIntercept -0.215802 0.026132\nX1 -0.151988 0.093336\nX2 0.626301 0.877282\n P-Value\nIntercept 0.123932\nX1 0.638339\nX2 0.000000\n" ], [ "reg = smf.ols(formula='Y ~ X2', data=pd.DataFrame({'X2':data['X2'],'Y':data['Y']})).fit()", "_____no_output_____" ], [ "sm.qqplot(reg.resid, line='45')\nplt.show()", "_____no_output_____" ], [ "sns.relplot(x='Residuals',y='Predicted',data=pd.DataFrame({'Residuals':reg.resid,'Predicted':reg.fittedvalues}))", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "from sklearn.svm import SVC\nfrom sklearn.preprocessing import StandardScaler\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.grid_search import GridSearchCV\nfrom sklearn import metrics\nimport numpy as np\nimport pandas as pd\nfrom matplotlib import pylab as plt\nimport re\n\n%matplotlib inline", "/Users/zytec/anaconda3/lib/python3.6/site-packages/sklearn/cross_validation.py:41: DeprecationWarning: This module was deprecated in version 0.18 in favor of the model_selection module into which all the refactored classes and functions are moved. Also note that the interface of the new CV iterators are different from that of this module. This module will be removed in 0.20.\n \"This module will be removed in 0.20.\", DeprecationWarning)\n/Users/zytec/anaconda3/lib/python3.6/site-packages/sklearn/grid_search.py:42: DeprecationWarning: This module was deprecated in version 0.18 in favor of the model_selection module into which all the refactored classes and functions are moved. This module will be removed in 0.20.\n DeprecationWarning)\n" ] ], [ [ "## 数据读取", "_____no_output_____" ] ], [ [ "df = pd.read_csv('data/train.csv')", "_____no_output_____" ], [ "df.head(10)", "_____no_output_____" ], [ "print('共有{}条数据.'.format(len(df)))\nprint('幸存者{}人.'.format(df.Survived.sum()))", "共有891条数据.\n幸存者342人.\n" ] ], [ [ "## 数据分析", "_____no_output_____" ], [ "#### Pclass 代表社会经济状况，数字越小状况越好 1 = upper, 2 = mid, 3 = lower", "_____no_output_____" ] ], [ [ "df.Pclass.value_counts(dropna=False)", "_____no_output_____" ], [ "df[['Pclass', 'Survived']].groupby(['Pclass']).mean().plot(kind='bar')", "_____no_output_____" ] ], [ [ "*社会等级越高，幸存率越高*", "_____no_output_____" ] ], [ [ "df['P1'] = (df.Pclass == 1).astype('int')\ndf['P2'] = (df.Pclass == 2).astype('int')\ndf['P3'] = (df.Pclass == 3).astype('int')\ndf.head()", "_____no_output_____" ] ], [ [ "#### Sex male时为1， female时为0", "_____no_output_____" ] ], [ [ "df.Sex.value_counts(dropna=False)", "_____no_output_____" ], [ "df[['Sex', 'Survived']].groupby(['Sex']).mean().plot(kind='bar')", "_____no_output_____" ] ], [ [ "*女性的幸存率比较高*", "_____no_output_____" ] ], [ [ "df.Sex.replace(['male', 'female'], [1, 0], inplace=True)\ndf.head()", "_____no_output_____" ] ], [ [ "#### Age 有缺失值，简单用平均值填充", "_____no_output_____" ] ], [ [ "df.Age.isnull().sum()", "_____no_output_____" ], [ "df.Age.fillna(df.Age.median(), inplace=True)", "_____no_output_____" ], [ "df['Age_categories'] = pd.cut(df['Age'], 5)", "_____no_output_____" ], [ "df[['Age_categories', 'Survived']].groupby(['Age_categories']).mean().plot(kind='bar')", "_____no_output_____" ] ], [ [ "*可见小孩的幸存率比较高*", "_____no_output_____" ], [ "#### Sibsp\t在船上的旁系亲属和配偶的数量", "_____no_output_____" ] ], [ [ "df.SibSp.isnull().sum()", "_____no_output_____" ] ], [ [ "#### Parch\t在船上的父母子女的数量", "_____no_output_____" ] ], [ [ "df.Parch.isnull().sum()", "_____no_output_____" ], [ "df['Family_size'] = df['SibSp'] + df['Parch']\ndf[['Family_size', 'Survived']].groupby(['Family_size']).mean().plot(kind='bar')", "_____no_output_____" ] ], [ [ "*亲属和配偶的数量和父母子女的数量合并为一个特征*", "_____no_output_____" ], [ "#### Fare\t船费", "_____no_output_____" ] ], [ [ "df.Fare.isnull().sum()", "_____no_output_____" ], [ "df['Fare_categories'] = pd.qcut(df['Fare'], 5)\ndf[['Fare_categories', 'Survived']].groupby(['Fare_categories']).mean().plot(kind='bar')", "_____no_output_____" ] ], [ [ "*票价越贵，乘客越有钱，幸存几率越大*", "_____no_output_____" ], [ "#### Embarked\t登船港口 C = Cherbourg, Q = Queenstown, S = Southampton", "_____no_output_____" ] ], [ [ "df.Embarked.value_counts(dropna=False)", "_____no_output_____" ], [ "df[['Embarked', 'Survived']].groupby(['Embarked']).mean().plot(kind='bar')", "_____no_output_____" ] ], [ [ "*可以看到在Cherbourg登陆的乘客幸存几率大*", "_____no_output_____" ] ], [ [ "df['E1'] = (df['Embarked'] == 'S').astype('int')\ndf['E2'] = (df['Embarked'] == 'C').astype('int')\ndf['E3'] = (df['Embarked'] == 'Q').astype('int')\ndf['E4'] = (df['Embarked'].isnull()).astype('int')\ndf.head()", "_____no_output_____" ] ], [ [ "#### 称谓，Mr、Miss、Master ...", "_____no_output_____" ] ], [ [ "def parse_title(name):\n match = re.match(r'\\w+\\,\\s(\\w+)\\.\\s', name)\n if match:\n return match.group(1)\n else:\n return np.nan", "_____no_output_____" ], [ "df['Title'] = df.Name.apply(parse_title)\ndf.Title.fillna('NoTitle', inplace=True)\ndf.Title.value_counts(dropna=False)", "_____no_output_____" ], [ "title_set = set(df.Title)\nfor i, title in enumerate(title_set):\n df['T'+str(i)] = (df.Title == title).astype('int')", "_____no_output_____" ], [ "df.head(3)", "_____no_output_____" ] ], [ [ "### 预处理后的数据", "_____no_output_____" ] ], [ [ "df_processed = df[['Survived', 'Sex', 'Age', 'Family_size', 'Fare', 'Pclass',\n 'E1', 'E2', 'E3', 'E4', 'T0', 'T1', 'T2', 'T3', 'T4', 'T5',\n 'T6', 'T7', 'T8', 'T9', 'T10', 'T11', 'T12', 'T13', 'T14']]\ndf_processed.head(3)", "_____no_output_____" ] ], [ [ "## 划分训练集和测试集", "_____no_output_____" ] ], [ [ "total_X = df_processed.iloc[:, 1:].values\ntotal_Y = df_processed.iloc[:, 0].values", "_____no_output_____" ], [ "train_X, test_X, train_Y, test_Y = train_test_split(total_X, total_Y, test_size=0.25)", "_____no_output_____" ] ], [ [ "## 数据标准化", "_____no_output_____" ] ], [ [ "X_scaler = StandardScaler()\ntrain_X_std = X_scaler.fit_transform(train_X)\ntest_X_std = X_scaler.transform(test_X)", "_____no_output_____" ] ], [ [ "### SVM超参数网格搜索", "_____no_output_____" ] ], [ [ "svm = SVC()\n\nparams = {\n 'C': np.logspace(1, (2 * np.random.rand()), 10),\n 'gamma':np.logspace(-4, (2 * np.random.rand()), 10)\n}\n\ngrid_search = GridSearchCV(svm, params, cv=3)\ngrid_search.fit(train_X_std, train_Y)", "_____no_output_____" ], [ "best_score = grid_search.best_score_", "_____no_output_____" ], [ "best_params = grid_search.best_params_\nC = best_params['C']\ngamma = best_params['gamma']\nC, gamma", "_____no_output_____" ], [ "grid_search.score(test_X_std, test_Y)", "_____no_output_____" ], [ "predicts = grid_search.predict(test_X_std)\nprint(metrics.classification_report(test_Y, predicts))", " precision recall f1-score support\n\n 0 0.78 0.95 0.85 129\n 1 0.89 0.63 0.74 94\n\navg / total 0.83 0.81 0.80 223\n\n" ] ], [ [ "## 最终预测", "_____no_output_____" ] ], [ [ "total_X_std = X_scaler.transform(df_processed.iloc[:, 1:].values)\ntotal_Y = df_processed.iloc[:, 0]", "_____no_output_____" ], [ "svm = SVC(C=C, gamma=gamma)\nsvm.fit(total_X_std, total_Y)", "_____no_output_____" ], [ "svm.score(total_X_std, total_Y)", "_____no_output_____" ], [ "test_df = pd.read_csv('data/test.csv')\ntest_df.head()", "_____no_output_____" ], [ "test_df.Sex.replace(['male', 'female'], [1, 0], inplace=True)\ntest_df.Age.fillna(df.Age.median(), inplace=True)\ntest_df['Family_size'] = test_df['SibSp'] + test_df['Parch']\ntest_df['P1'] = (df.Pclass == 1).astype('int')\ntest_df['P2'] = (df.Pclass == 2).astype('int')\ntest_df['P3'] = (df.Pclass == 3).astype('int')\ntest_df['E1'] = (df['Embarked'] == 'S').astype('int')\ntest_df['E2'] = (df['Embarked'] == 'C').astype('int')\ntest_df['E3'] = (df['Embarked'] == 'Q').astype('int')\ntest_df['E4'] = (df['Embarked'].isnull()).astype('int')\ntest_df['Title'] = df.Name.apply(parse_title)\ntest_df.Title.fillna('NoTitle', inplace=True)\ntitle_set = set(df.Title)\nfor i, title in enumerate(title_set):\n test_df['T'+str(i)] = (test_df.Title == title).astype('int')\n\ntest_df.head()", "_____no_output_____" ], [ "test_df.isnull().sum()", "_____no_output_____" ], [ "test_df.Fare.fillna(df.Fare.median(), inplace=True)", "_____no_output_____" ], [ "test_df_processed = test_df[['Sex', 'Age', 'Family_size', 'Fare', 'Pclass',\n 'E1', 'E2', 'E3', 'E4', 'T0', 'T1', 'T2', 'T3', 'T4', 'T5',\n 'T6', 'T7', 'T8', 'T9', 'T10', 'T11', 'T12', 'T13', 'T14']]\ntest_df_processed.head(3)", "_____no_output_____" ], [ "final_X = test_df_processed.values\nfinal_X_std = X_scaler.transform(final_X)", "_____no_output_____" ] ], [ [ "### 预测", "_____no_output_____" ] ], [ [ "predicts = svm.predict(final_X_std)", "_____no_output_____" ], [ "ids = test_df['PassengerId'].values", "_____no_output_____" ], [ "result = pd.DataFrame({\n 'PassengerId': ids,\n 'Survived': predicts\n})", "_____no_output_____" ], [ "result.head()", "_____no_output_____" ], [ "result.to_csv('./2018-8-23_6.csv', index=False)", "_____no_output_____" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# Sometimes we need an immutable collection of related constant elements\nimport enum\n\nclass Color(enum.Enum):\n red = 1\n green = 2\n blue = 3\n\nclass Status(enum.Enum):\n PENDING = 'PENDING'\n RUNNING = 'RUNNING'\n COMPLETE = 'COMPLETE'", "_____no_output_____" ], [ "type(Color.red)", "_____no_output_____" ], [ "isinstance(Color.red, Color)", "_____no_output_____" ], [ "Color.red.name, Color.red.value", "_____no_output_____" ], [ "list(Status)", "_____no_output_____" ], [ "for status in list(Status):\n print(status)", "Status.PENDING\nStatus.RUNNING\nStatus.COMPLETE\n" ], [ "type(Status)", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code" ] ]

[ "Xnet", "X11" ]

[ "Xnet", "X11" ]

[ "Xnet", "X11" ]

[ [ [ "def grad(x):\n h=0.41\n while x <=1:\n return -1\n while x>1 and x<(1+h):\n return 1\n while x>(1+h) and x<(1+2*h):\n return -1", "_____no_output_____" ], [ "h=0.41\nlr=0.3\nbeta1=0.9\nbeta2=0.999\nepislon=0\nm=[0]\nv = [0]\nt = 0\nxt=0", "_____no_output_____" ], [ "while 1:\n t+=1\n gx=grad(xt)\n m.append(beta1*m[t-1]+(1-beta1)*gx)\n v.append(beta2*v[t-1]+(1-beta2)*gx*gx)\n m_hat=m[t] / (1 - beta1 ** t)\n v_hat=v[t] / (1 - beta2 ** t)\n xt -= lr *m_hat / (v_hat ** 0.5 + epislon)\n print('t=',t,'xt=',xt)", "t= 1 xt= 0.3\nt= 2 xt= 0.5999999999999979\nt= 3 xt= 0.8999999999999977\nt= 4 xt= 1.199999999999996\nt= 5 xt= 1.353483431418032\nt= 6 xt= 1.4101842951299657\nt= 7 xt= 1.5135207138708162\nt= 8 xt= 1.6513878189412725\nt= 9 xt= 1.8157221649197492\nt= 10 xt= 2.000885800510925\n" ] ] ]

[ "code" ]

[ [ "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "<a href=\"https://colab.research.google.com/github/Tessellate-Imaging/monk_v1/blob/master/study_roadmaps/3_image_processing_deep_learning_roadmap/3_deep_learning_advanced/1_Blocks%20in%20Deep%20Learning%20Networks/8)%20Resnet%20V2%20Bottleneck%20Block%20(Type%20-%202).ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>", "_____no_output_____" ], [ "# Goals\n\n### 1. Learn to implement Resnet V2 Bottleneck Block (Type - 1) using monk\n - Monk's Keras\n - Monk's Pytorch\n - Monk's Mxnet\n \n### 2. Use network Monk's debugger to create complex blocks \n\n\n### 3. Understand how syntactically different it is to implement the same using\n - Traditional Keras\n - Traditional Pytorch\n - Traditional Mxnet", "_____no_output_____" ], [ "# Resnet V2 Bottleneck Block - Type 1\n \n - Note: The block structure can have variations too, this is just an example", "_____no_output_____" ] ], [ [ "from IPython.display import Image\nImage(filename='imgs/resnet_v2_bottleneck_without_downsample.png')", "_____no_output_____" ] ], [ [ "# Table of contents\n\n[1. Install Monk](#1)\n\n\n[2. Block basic Information](#2)\n\n - [2.1) Visual structure](#2-1)\n \n - [2.2) Layers in Branches](#2-2)\n\n\n[3) Creating Block using monk visual debugger](#3)\n\n - [3.1) Create the first branch](#3-1)\n\n - [3.2) Create the second branch](#3-2)\n \n - [3.3) Merge the branches](#3-3)\n \n - [3.4) Debug the merged network](#3-4)\n \n - [3.5) Compile the network](#3-5)\n \n - [3.6) Visualize the network](#3-6)\n \n - [3.7) Run data through the network](#3-7)\n \n \n[4) Creating Block Using MONK one line API call](#4)\n\n - [Mxnet Backend](#4-1)\n \n - [Pytorch Backend](#4-2)\n \n - [Keras Backend](#4-3)\n \n \n \n[5) Appendix](#5)\n\n - [Study Material](#5-1)\n \n - [Creating block using traditional Mxnet](#5-2)\n \n - [Creating block using traditional Pytorch](#5-3)\n \n - [Creating block using traditional Keras](#5-4)\n ", "_____no_output_____" ], [ "<a id='0'></a>\n# Install Monk", "_____no_output_____" ], [ "## Using pip (Recommended)\n\n - colab (gpu) \n - All bakcends: `pip install -U monk-colab`\n \n\n - kaggle (gpu) \n - All backends: `pip install -U monk-kaggle`\n \n\n - cuda 10.2\t\n - All backends: `pip install -U monk-cuda102`\n - Gluon bakcned: `pip install -U monk-gluon-cuda102`\n\t - Pytorch backend: `pip install -U monk-pytorch-cuda102`\n - Keras backend: `pip install -U monk-keras-cuda102`\n \n\n - cuda 10.1\t\n - All backend: `pip install -U monk-cuda101`\n\t - Gluon bakcned: `pip install -U monk-gluon-cuda101`\n\t - Pytorch backend: `pip install -U monk-pytorch-cuda101`\n\t - Keras backend: `pip install -U monk-keras-cuda101`\n \n\n - cuda 10.0\t\n - All backend: `pip install -U monk-cuda100`\n\t - Gluon bakcned: `pip install -U monk-gluon-cuda100`\n\t - Pytorch backend: `pip install -U monk-pytorch-cuda100`\n\t - Keras backend: `pip install -U monk-keras-cuda100`\n \n\n - cuda 9.2\t\n - All backend: `pip install -U monk-cuda92`\n\t - Gluon bakcned: `pip install -U monk-gluon-cuda92`\n\t - Pytorch backend: `pip install -U monk-pytorch-cuda92`\n\t - Keras backend: `pip install -U monk-keras-cuda92`\n \n\n - cuda 9.0\t\n - All backend: `pip install -U monk-cuda90`\n\t - Gluon bakcned: `pip install -U monk-gluon-cuda90`\n\t - Pytorch backend: `pip install -U monk-pytorch-cuda90`\n\t - Keras backend: `pip install -U monk-keras-cuda90`\n \n\n - cpu \t\t\n - All backend: `pip install -U monk-cpu`\n\t - Gluon bakcned: `pip install -U monk-gluon-cpu`\n\t - Pytorch backend: `pip install -U monk-pytorch-cpu`\n\t - Keras backend: `pip install -U monk-keras-cpu`", "_____no_output_____" ], [ "## Install Monk Manually (Not recommended)\n \n### Step 1: Clone the library\n - git clone https://github.com/Tessellate-Imaging/monk_v1.git\n \n \n \n \n### Step 2: Install requirements \n - Linux\n - Cuda 9.0\n - `cd monk_v1/installation/Linux && pip install -r requirements_cu90.txt`\n - Cuda 9.2\n - `cd monk_v1/installation/Linux && pip install -r requirements_cu92.txt`\n - Cuda 10.0\n - `cd monk_v1/installation/Linux && pip install -r requirements_cu100.txt`\n - Cuda 10.1\n - `cd monk_v1/installation/Linux && pip install -r requirements_cu101.txt`\n - Cuda 10.2\n - `cd monk_v1/installation/Linux && pip install -r requirements_cu102.txt`\n - CPU (Non gpu system)\n - `cd monk_v1/installation/Linux && pip install -r requirements_cpu.txt`\n \n \n - Windows\n - Cuda 9.0 (Experimental support)\n - `cd monk_v1/installation/Windows && pip install -r requirements_cu90.txt`\n - Cuda 9.2 (Experimental support)\n - `cd monk_v1/installation/Windows && pip install -r requirements_cu92.txt`\n - Cuda 10.0 (Experimental support)\n - `cd monk_v1/installation/Windows && pip install -r requirements_cu100.txt`\n - Cuda 10.1 (Experimental support)\n - `cd monk_v1/installation/Windows && pip install -r requirements_cu101.txt`\n - Cuda 10.2 (Experimental support)\n - `cd monk_v1/installation/Windows && pip install -r requirements_cu102.txt`\n - CPU (Non gpu system)\n - `cd monk_v1/installation/Windows && pip install -r requirements_cpu.txt`\n \n \n - Mac\n - CPU (Non gpu system)\n - `cd monk_v1/installation/Mac && pip install -r requirements_cpu.txt`\n \n \n - Misc\n - Colab (GPU)\n - `cd monk_v1/installation/Misc && pip install -r requirements_colab.txt`\n - Kaggle (GPU)\n - `cd monk_v1/installation/Misc && pip install -r requirements_kaggle.txt`\n \n \n \n### Step 3: Add to system path (Required for every terminal or kernel run)\n - `import sys`\n - `sys.path.append(\"monk_v1/\");`", "_____no_output_____" ], [ "# Imports", "_____no_output_____" ] ], [ [ "# Common\nimport numpy as np\nimport math\nimport netron\nfrom collections import OrderedDict\nfrom functools import partial", "_____no_output_____" ], [ "#Using mxnet-gluon backend \n\n# When installed using pip\nfrom monk.gluon_prototype import prototype\n\n\n# When installed manually (Uncomment the following)\n#import os\n#import sys\n#sys.path.append(\"monk_v1/\");\n#sys.path.append(\"monk_v1/monk/\");\n#from monk.gluon_prototype import prototype", "_____no_output_____" ] ], [ [ "<a id='2'></a>\n# Block Information", "_____no_output_____" ], [ "<a id='2_1'></a>\n## Visual structure", "_____no_output_____" ] ], [ [ "from IPython.display import Image\nImage(filename='imgs/resnet_v2_bottleneck_without_downsample.png')", "_____no_output_____" ] ], [ [ "<a id='2_2'></a>\n## Layers in Branches\n\n - Number of branches: 2\n \n \n - Common Elements\n - batchnorm -> relu\n \n \n - Branch 1\n - identity\n \n \n - Branch 2\n - conv_1x1 -> batchnorm -> relu -> conv_3x3 -> batchnorm -> relu -> conv1x1\n \n \n - Branches merged using\n - Elementwise addition\n \n \n(See Appendix to read blogs on resnets)", "_____no_output_____" ], [ "<a id='3'></a>\n# Creating Block using monk debugger", "_____no_output_____" ] ], [ [ "# Imports and setup a project\n# To use pytorch backend - replace gluon_prototype with pytorch_prototype\n# To use keras backend - replace gluon_prototype with keras_prototype\n\nfrom monk.gluon_prototype import prototype\n\n# Create a sample project\ngtf = prototype(verbose=1);\ngtf.Prototype(\"sample-project-1\", \"sample-experiment-1\");", "Mxnet Version: 1.5.1\n\nExperiment Details\n Project: sample-project-1\n Experiment: sample-experiment-1\n Dir: /home/abhi/Desktop/Work/tess_tool/gui/v0.3/finetune_models/Organization/development/v5.0_blocks/study_roadmap/blocks/workspace/sample-project-1/sample-experiment-1/\n\n" ] ], [ [ "<a id='3-1'></a>\n## Create the first branch", "_____no_output_____" ] ], [ [ "def first_branch():\n network = [];\n network.append(gtf.identity());\n return network;", "_____no_output_____" ], [ "# Debug the branch\nbranch_1 = first_branch()\nnetwork = [];\nnetwork.append(branch_1);\ngtf.debug_custom_model_design(network);", "_____no_output_____" ] ], [ [ "<a id='3-2'></a>\n## Create the second branch", "_____no_output_____" ] ], [ [ "def second_branch(output_channels=128, stride=1):\n network = [];\n \n # Bottleneck convolution \n network.append(gtf.convolution(output_channels=output_channels//4, kernel_size=1, stride=stride));\n network.append(gtf.batch_normalization());\n network.append(gtf.relu());\n \n #Bottleneck convolution\n network.append(gtf.convolution(output_channels=output_channels//4, kernel_size=1, stride=stride));\n network.append(gtf.batch_normalization());\n network.append(gtf.relu());\n \n #Normal convolution\n network.append(gtf.convolution(output_channels=output_channels, kernel_size=1, stride=1));\n return network;", "_____no_output_____" ], [ "# Debug the branch\nbranch_2 = second_branch(output_channels=128, stride=1)\nnetwork = [];\nnetwork.append(branch_2);\ngtf.debug_custom_model_design(network);", "_____no_output_____" ] ], [ [ "<a id='3-3'></a>\n## Merge the branches", "_____no_output_____" ] ], [ [ "\ndef final_block(output_channels=128, stride=1):\n network = [];\n \n #Common Elements\n network.append(gtf.batch_normalization());\n network.append(gtf.relu());\n \n #Create subnetwork and add branches\n subnetwork = [];\n branch_1 = first_branch()\n branch_2 = second_branch(output_channels=output_channels, stride=stride)\n subnetwork.append(branch_1);\n subnetwork.append(branch_2);\n \n # Add merging element\n subnetwork.append(gtf.add());\n \n # Add the subnetwork\n network.append(subnetwork)\n return network;", "_____no_output_____" ] ], [ [ "<a id='3-4'></a>\n## Debug the merged network", "_____no_output_____" ] ], [ [ "final = final_block(output_channels=64, stride=1)\nnetwork = [];\nnetwork.append(final);\ngtf.debug_custom_model_design(network);", "_____no_output_____" ] ], [ [ "<a id='3-5'></a>\n## Compile the network", "_____no_output_____" ] ], [ [ "gtf.Compile_Network(network, data_shape=(64, 224, 224), use_gpu=False);", "Model Details\n Loading pretrained model\n Model Loaded on device\n Model name: Custom Model\n Num of potentially trainable layers: 6\n Num of actual trainable layers: 6\n\n" ] ], [ [ "<a id='3-6'></a>\n## Run data through the network", "_____no_output_____" ] ], [ [ "import mxnet as mx\nx = np.zeros((1, 64, 224, 224));\nx = mx.nd.array(x);\ny = gtf.system_dict[\"local\"][\"model\"].forward(x);\nprint(x.shape, y.shape)", "(1, 64, 224, 224) (1, 64, 224, 224)\n" ] ], [ [ "<a id='3-7'></a>\n## Visualize network using netron", "_____no_output_____" ] ], [ [ "gtf.Visualize_With_Netron(data_shape=(64, 224, 224))", "Using Netron To Visualize\nNot compatible on kaggle\nCompatible only for Jupyter Notebooks\n\nStopping http://localhost:8080\nServing 'model-symbol.json' at http://localhost:8080\n" ] ], [ [ "<a id='4'></a>\n# Creating Using MONK LOW code API", "_____no_output_____" ], [ "<a id='4-1'></a>\n## Mxnet backend", "_____no_output_____" ] ], [ [ "from monk.gluon_prototype import prototype\ngtf = prototype(verbose=1);\ngtf.Prototype(\"sample-project-1\", \"sample-experiment-1\");\n\n\nnetwork = [];\n\n# Single line addition of blocks\nnetwork.append(gtf.resnet_v2_bottleneck_block(output_channels=64, downsample=False));\n\n\ngtf.Compile_Network(network, data_shape=(64, 224, 224), use_gpu=False);\n", "Mxnet Version: 1.5.1\n\nExperiment Details\n Project: sample-project-1\n Experiment: sample-experiment-1\n Dir: /home/abhi/Desktop/Work/tess_tool/gui/v0.3/finetune_models/Organization/development/v5.0_blocks/study_roadmap/blocks/workspace/sample-project-1/sample-experiment-1/\n\nModel Details\n Loading pretrained model\n Model Loaded on device\n Model name: Custom Model\n Num of potentially trainable layers: 6\n Num of actual trainable layers: 6\n\n" ] ], [ [ "<a id='4-2'></a>\n## Pytorch backend\n\n - Only the import changes", "_____no_output_____" ] ], [ [ "#Change gluon_prototype to pytorch_prototype\nfrom monk.pytorch_prototype import prototype\n\n\ngtf = prototype(verbose=1);\ngtf.Prototype(\"sample-project-1\", \"sample-experiment-1\");\n\n\nnetwork = [];\n\n# Single line addition of blocks\nnetwork.append(gtf.resnet_v2_bottleneck_block(output_channels=64, downsample=False));\n\n\ngtf.Compile_Network(network, data_shape=(64, 224, 224), use_gpu=False);", "Pytorch Version: 1.2.0\n\nExperiment Details\n Project: sample-project-1\n Experiment: sample-experiment-1\n Dir: /home/abhi/Desktop/Work/tess_tool/gui/v0.3/finetune_models/Organization/development/v5.0_blocks/study_roadmap/blocks/workspace/sample-project-1/sample-experiment-1/\n\nModel Details\n Loading pretrained model\n Model Loaded on device\n Model name: Custom Model\n Num layers in model: 6\n Num trainable layers: 6\n\n" ] ], [ [ "<a id='4-3'></a>\n## Keras backend\n\n - Only the import changes", "_____no_output_____" ] ], [ [ "#Change gluon_prototype to keras_prototype\nfrom monk.keras_prototype import prototype\n\n\ngtf = prototype(verbose=1);\ngtf.Prototype(\"sample-project-1\", \"sample-experiment-1\");\n\n\nnetwork = [];\n\n# Single line addition of blocks\nnetwork.append(gtf.resnet_v2_bottleneck_block(output_channels=64, downsample=False));\n\n\ngtf.Compile_Network(network, data_shape=(64, 224, 224), use_gpu=False);", "Keras Version: 2.2.5\nTensorflow Version: 1.12.0\n\nExperiment Details\n Project: sample-project-1\n Experiment: sample-experiment-1\n Dir: /home/abhi/Desktop/Work/tess_tool/gui/v0.3/finetune_models/Organization/development/v5.0_blocks/study_roadmap/blocks/workspace/sample-project-1/sample-experiment-1/\n\nModel Details\n Loading pretrained model\n Model Loaded on device\n Model name: Custom Model\n Num layers in model: 11\n Num trainable layers: 10\n\n" ] ], [ [ "<a id='5'></a>\n# Appendix", "_____no_output_____" ], [ "<a id='5-1'></a>\n## Study links\n - https://towardsdatascience.com/residual-blocks-building-blocks-of-resnet-fd90ca15d6ec\n - https://medium.com/@MaheshNKhatri/resnet-block-explanation-with-a-terminology-deep-dive-989e15e3d691\n - https://medium.com/analytics-vidhya/understanding-and-implementation-of-residual-networks-resnets-b80f9a507b9c\n - https://hackernoon.com/resnet-block-level-design-with-deep-learning-studio-part-1-727c6f4927ac", "_____no_output_____" ], [ "<a id='5-2'></a>\n## Creating block using traditional Mxnet\n\n - Code credits - https://mxnet.incubator.apache.org/", "_____no_output_____" ] ], [ [ "# Traditional-Mxnet-gluon\nimport mxnet as mx\nfrom mxnet.gluon import nn\nfrom mxnet.gluon.nn import HybridBlock, BatchNorm\nfrom mxnet.gluon.contrib.nn import HybridConcurrent, Identity\nfrom mxnet import gluon, init, nd", "_____no_output_____" ], [ "\ndef _conv3x3(channels, stride, in_channels):\n return nn.Conv2D(channels, kernel_size=3, strides=stride, padding=1,\n use_bias=False, in_channels=in_channels)\n\nclass ResnetBlockV1(HybridBlock):\n def __init__(self, channels, stride, in_channels=0, **kwargs):\n super(ResnetBlockV1, self).__init__(**kwargs)\n \n #Common Elements\n self.bn0 = nn.BatchNorm();\n self.relu0 = nn.Activation('relu');\n \n #Branch - 1\n #Identity\n \n \n # Branch - 2\n self.body = nn.HybridSequential(prefix='')\n self.body.add(nn.Conv2D(channels//4, kernel_size=1, strides=stride,\n use_bias=False, in_channels=in_channels))\n self.body.add(nn.BatchNorm())\n self.body.add(nn.Activation('relu'))\n self.body.add(_conv3x3(channels//4, stride, in_channels))\n self.body.add(nn.BatchNorm())\n self.body.add(nn.Activation('relu'))\n self.body.add(nn.Conv2D(channels, kernel_size=1, strides=stride,\n use_bias=False, in_channels=in_channels))\n \n\n def hybrid_forward(self, F, x):\n x = self.bn0(x);\n x = self.relu0(x);\n \n residual = x\n x = self.body(x)\n x = residual+x\n return x", "_____no_output_____" ], [ "# Invoke the block\nblock = ResnetBlockV1(64, 1)\n\n# Initialize network and load block on machine\nctx = [mx.cpu()];\nblock.initialize(init.Xavier(), ctx = ctx);\nblock.collect_params().reset_ctx(ctx)\nblock.hybridize()\n\n# Run data through network\nx = np.zeros((1, 64, 224, 224));\nx = mx.nd.array(x);\ny = block.forward(x);\nprint(x.shape, y.shape)\n\n\n# Export Model to Load on Netron\nblock.export(\"final\", epoch=0);\n\nnetron.start(\"final-symbol.json\", port=8082)", "(1, 64, 224, 224) (1, 64, 224, 224)\nServing 'final-symbol.json' at http://localhost:8082\n" ] ], [ [ "<a id='5-3'></a>\n## Creating block using traditional Pytorch\n\n - Code credits - https://pytorch.org/", "_____no_output_____" ] ], [ [ "# Traiditional-Pytorch\nimport torch\nfrom torch import nn\nfrom torch.jit.annotations import List\nimport torch.nn.functional as F", "_____no_output_____" ], [ "def conv3x3(in_planes, out_planes, stride=1, groups=1, dilation=1):\n \"\"\"3x3 convolution with padding\"\"\"\n return nn.Conv2d(in_planes, out_planes, kernel_size=3, stride=stride,\n padding=dilation, groups=groups, bias=False, dilation=dilation)\n\n\ndef conv1x1(in_planes, out_planes, stride=1):\n \"\"\"1x1 convolution\"\"\"\n return nn.Conv2d(in_planes, out_planes, kernel_size=1, stride=stride, bias=False)\n\n\nclass ResnetBottleNeckBlock(nn.Module):\n expansion = 1\n __constants__ = ['downsample']\n\n def __init__(self, inplanes, planes, stride=1, groups=1,\n base_width=64, dilation=1, norm_layer=None):\n super(ResnetBottleNeckBlock, self).__init__()\n \n norm_layer = nn.BatchNorm2d\n \n #Common elements\n self.bn0 = norm_layer(inplanes);\n self.relu0 = nn.ReLU(inplace=True);\n \n \n # Branch - 1\n #Identity\n \n \n # Branch - 2\n self.conv1 = conv1x1(inplanes, planes//4, stride)\n self.bn1 = norm_layer(planes//4)\n self.relu1 = nn.ReLU(inplace=True)\n self.conv2 = conv3x3(planes//4, planes//4, stride)\n self.bn2 = norm_layer(planes//4)\n self.relu2 = nn.ReLU(inplace=True)\n \n self.conv3 = conv1x1(planes//4, planes)\n \n self.stride = stride\n\n self.relu = nn.ReLU(inplace=True)\n \n def forward(self, x):\n x = self.bn0(x);\n x = self.relu0(x); \n \n identity = x\n\n out = self.conv1(x)\n out = self.bn1(out)\n out = self.relu1(out)\n \n out = self.conv2(out)\n out = self.bn2(out)\n out = self.relu2(out)\n\n out = self.conv3(out)\n\n out += identity\n\n return out", "_____no_output_____" ], [ "# Invoke the block\nblock = ResnetBottleNeckBlock(64, 64, stride=1);\n\n# Initialize network and load block on machine\nlayers = []\nlayers.append(block);\nnet = nn.Sequential(*layers);\n\n# Run data through network\nx = torch.randn(1, 64, 224, 224)\ny = net(x)\nprint(x.shape, y.shape);\n\n# Export Model to Load on Netron\ntorch.onnx.export(net, # model being run\n x, # model input (or a tuple for multiple inputs)\n \"model.onnx\", # where to save the model (can be a file or file-like object)\n export_params=True, # store the trained parameter weights inside the model file\n opset_version=10, # the ONNX version to export the model to\n do_constant_folding=True, # whether to execute constant folding for optimization\n input_names = ['input'], # the model's input names\n output_names = ['output'], # the model's output names\n dynamic_axes={'input' : {0 : 'batch_size'}, # variable lenght axes\n 'output' : {0 : 'batch_size'}})\nnetron.start('model.onnx', port=9998);\n", "torch.Size([1, 64, 224, 224]) torch.Size([1, 64, 224, 224])\nServing 'model.onnx' at http://localhost:9998\n" ] ], [ [ "<a id='5-4'></a>\n## Creating block using traditional Keras\n\n - Code credits: https://keras.io/", "_____no_output_____" ] ], [ [ "# Traditional-Keras\nimport keras\nimport keras.layers as kla\nimport keras.models as kmo\nimport tensorflow as tf\nfrom keras.models import Model\nbackend = 'channels_last'\nfrom keras import layers", "_____no_output_____" ], [ "def resnet_conv_block(input_tensor,\n kernel_size,\n filters,\n stage,\n block,\n strides=(1, 1)):\n\n filters1, filters2, filters3 = filters\n bn_axis = 3\n \n conv_name_base = 'res' + str(stage) + block + '_branch'\n bn_name_base = 'bn' + str(stage) + block + '_branch'\n\n \n #Common Elements\n start = layers.BatchNormalization(axis=bn_axis, name=bn_name_base + '0a')(input_tensor)\n start = layers.Activation('relu')(start)\n \n # Branch - 1\n # Identity\n shortcut = start\n \n \n # Branch - 2\n x = layers.Conv2D(filters1, (1, 1), strides=strides,\n kernel_initializer='he_normal',\n name=conv_name_base + '2a')(start)\n x = layers.BatchNormalization(axis=bn_axis, name=bn_name_base + '2a')(x)\n x = layers.Activation('relu')(x)\n \n x = layers.Conv2D(filters2, (3, 3), strides=strides,\n kernel_initializer='he_normal',\n name=conv_name_base + '2b', padding=\"same\")(x)\n x = layers.BatchNormalization(axis=bn_axis, name=bn_name_base + '2b')(x)\n x = layers.Activation('relu')(x)\n\n x = layers.Conv2D(filters3, (1, 1),\n kernel_initializer='he_normal',\n name=conv_name_base + '2c')(x);\n\n \n x = layers.add([x, shortcut])\n x = layers.Activation('relu')(x)\n return x\n\n\ndef create_model(input_shape, kernel_size, filters, stage, block):\n img_input = layers.Input(shape=input_shape);\n x = resnet_conv_block(img_input, kernel_size, filters, stage, block) \n return Model(img_input, x);", "_____no_output_____" ], [ "# Invoke the block\nkernel_size=3;\nfilters=[16, 16, 64];\ninput_shape=(224, 224, 64);\nmodel = create_model(input_shape, kernel_size, filters, 0, \"0\");\n\n# Run data through network\nx = tf.placeholder(tf.float32, shape=(1, 224, 224, 64))\ny = model(x)\nprint(x.shape, y.shape)\n\n# Export Model to Load on Netron\nmodel.save(\"final.h5\");\nnetron.start(\"final.h5\", port=8082)", "(1, 224, 224, 64) (1, 224, 224, 64)\n\nStopping http://localhost:8082\nServing 'final.h5' at http://localhost:8082\n" ] ], [ [ "# Goals Completed\n\n### 1. Learn to implement Resnet V2 Bottleneck Block (Type - 1) using monk\n - Monk's Keras\n - Monk's Pytorch\n - Monk's Mxnet\n \n### 2. Use network Monk's debugger to create complex blocks \n\n\n### 3. Understand how syntactically different it is to implement the same using\n - Traditional Keras\n - Traditional Pytorch\n - Traditional Mxnet", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import pandas as pd", "_____no_output_____" ], [ "file_dir = \"./Resources/\"", "_____no_output_____" ], [ "kaggle_metadata = pd.read_csv(f'{file_dir}movies_metadata.csv', low_memory=False)\nratings = pd.read_csv(f'{file_dir}ratings.csv')", "_____no_output_____" ], [ "kaggle_metadata.sample(n=5)", "_____no_output_____" ], [ "ratings.sample(n=5)", "_____no_output_____" ], [ "kaggle_metadata.dtypes", "_____no_output_____" ], [ "kaggle_metadata['adult'].value_counts()", "_____no_output_____" ], [ "kaggle_metadata[~kaggle_metadata['adult'].isin(['True','False'])]", "_____no_output_____" ], [ "kaggle_metadata = kaggle_metadata[kaggle_metadata['adult'] == 'False'].drop('adult',axis='columns')", "_____no_output_____" ], [ "kaggle_metadata['video'].value_counts()", "_____no_output_____" ], [ "kaggle_metadata['video'] == 'True'", "_____no_output_____" ], [ "kaggle_metadata['video'] = kaggle_metadata['video'] == 'True'", "_____no_output_____" ], [ "kaggle_metadata['budget'] = kaggle_metadata['budget'].astype(int)\nkaggle_metadata['id'] = pd.to_numeric(kaggle_metadata['id'], errors='raise')\nkaggle_metadata['popularity'] = pd.to_numeric(kaggle_metadata['popularity'], errors='raise')", "_____no_output_____" ], [ "kaggle_metadata['release_date'] = pd.to_datetime(kaggle_metadata['release_date'])", "_____no_output_____" ], [ "ratings.info(null_counts=True)", "<ipython-input-15-ed915c4d3989>:1: FutureWarning: null_counts is deprecated. Use show_counts instead\n ratings.info(null_counts=True)\n" ], [ "pd.to_datetime(ratings['timestamp'], unit='s')", "_____no_output_____" ], [ "ratings['timestamp'] = pd.to_datetime(ratings['timestamp'], unit='s')", "_____no_output_____" ], [ "pd.options.display.float_format = '{:20,.2f}'.format\nratings['rating'].plot(kind='hist')\nratings['rating'].describe()", "_____no_output_____" ], [ "# Competing data:\n# Wiki Movielens Resolution\n#--------------------------------------------------------------------------\n# title_wiki title_kaggle\n# running_time runtime\n# budget_wiki budget_kaggle\n# box_office revenue\n# release_date_wiki release_date_kaggle\n# Language original_language\n# Production company(s) production_companies", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "import pandas as pd\nimport geopandas as gpd\nimport matplotlib.pyplot as plt\nimport numpy as np\nfrom pyproj import CRS\nimport pathlib\nfrom pathlib import Path\nfrom shapely import wkt\nfrom tqdm import tqdm\n\nimport math\nimport codecs\n\nfrom shapely import wkt\n\nimport folium\nfrom folium import features\nfrom folium import plugins\n\nimport gzip\nfrom xml.etree.ElementTree import Element, SubElement, Comment, tostring\nimport xml.etree.ElementTree as ET\n\n# to read the excel \nfrom openpyxl import load_workbook\nfrom openpyxl import Workbook\n\n# import folium\nfrom shapely.geometry import LineString, MultiLineString\nimport branca.colormap as cmp\nfrom folium.plugins import Search\n\nfrom tqdm import tqdm\nimport time\n\nimport datetime\nfrom datetime import timedelta\n\n# set the working directory\nBASE_DIR = Path.cwd()\n", "_____no_output_____" ], [ "BASE_DIR", "_____no_output_____" ], [ "# save as geojson\ndef get_foldercreation_inf():\n fname = pathlib.Path(\"../SF_all_trips/sf-tscore-all-trips-20PCsample-updatedRideHailFleet-updatedParking__etg/ITERS/it.4/4.linkstats.csv.gz\")\n assert fname.exists(), f'No such file: {fname}' # check that the file exists\n ctime = datetime.datetime.fromtimestamp(fname.stat().st_ctime)\n return ctime\n# return ctime.strftime(\"%Y-%m-%d\")\n\ndef get_dataframe(_time):\n # linkstats file\n linkstats = pd.read_csv(\"../SF_all_trips/sf-tscore-all-trips-20PCsample-updatedRideHailFleet-updatedParking__etg/ITERS/it.4/4.linkstats.csv.gz\", compression=\"gzip\", low_memory=False)\n time = int(_time)\n linkstats = linkstats[linkstats[\"hour\"]==(time)].copy()\n linkstats=linkstats.add_prefix(\"linkstats_\")\n linkstats.rename(columns={('linkstats_link'): 'id'}, inplace=True)\n linkstats[\"id\"] = linkstats[\"id\"].astype('string')\n date_time = get_foldercreation_inf()\n if int(_time)<24:\n date_time = date_time.strftime(\"%Y-%m-%d\")\n time_stamp = f'{int(_time):02d}'\n linkstats[\"date_time\"] = (date_time + \" \" + \"{}:00:00\".format(f'{int(_time):02d}'))\n else:\n date_time = get_foldercreation_inf() + datetime.timedelta(days=1)\n date_time = date_time.strftime(\"%Y-%m-%d\")\n new_time = int(_time) - 24\n linkstats[\"date_time\"] = (date_time + \" \" + \"{}:00:00\".format(f'{abs(int(new_time)):02d}')) \n \n return linkstats\n\n\n# read the road network\nsf_roadnetwork = gpd.read_file(BASE_DIR.parent.joinpath( 'Network',\"sfNetwork.geojsonl\"))\nsf_roadnetwork = sf_roadnetwork[[\"id\",\"modes\",\"length\",\"lanes\",\"from\",\"to\",\"capacity\",\"geometry\"]]\nsftimevariantnetwork =pd.DataFrame()\n\n\nfor time_hour in tqdm(range(0,30)):\n # get the hour and filter the linkstat file\n linkstats = get_dataframe(str(time_hour))\n # merge with featureclass of SF data\n comparision_network = sf_roadnetwork.merge(linkstats,on=\"id\").copy()\n \n # calculate the freespeed (mph), congested speed (mph), ratio (congestedsped/freespeed)\n # linkstats\n comparision_network[\"linkstats_freespeed_mph\"] = comparision_network[\"linkstats_freespeed\"]*2.23694\n comparision_network[\"linkstats_congspd_mph\"] = (comparision_network[\"linkstats_length\"]/comparision_network[\"linkstats_traveltime\"])*2.23694\n comparision_network[\"linkstats_ratio\"] = comparision_network[\"linkstats_congspd_mph\"] / comparision_network[\"linkstats_freespeed_mph\"] \n comparision_network[\"linkstats_vc_ratio\"] = comparision_network[\"linkstats_volume\"]*5 / comparision_network[\"capacity\"]\n \n if int(time_hour)==0:\n sftimevariantnetwork = comparision_network.copy()\n else:\n sftimevariantnetwork = pd.concat([sftimevariantnetwork,comparision_network], ignore_index=True)\n\n# lastly, export the network\n# sftimevariantnetwork.to_file(BASE_DIR.parent.joinpath(\"exported\", (\"sf_timevariantnetwork.geojson\")), driver='GeoJSON')", "100%|██████████████████████████████████████████████████████████████████████████████████| 30/30 [03:50<00:00, 7.67s/it]\n" ], [ "linkstats.head()", "_____no_output_____" ], [ "# sftimevariantnetwork.to_csv(BASE_DIR.parent.joinpath(\"exported\", (\"sf_timevariantnetwork.csv\")))", "_____no_output_____" ], [ "# read the road network, incase it is already saved in the geojson file\n# sftimevariantnetwork = gpd.read_file(BASE_DIR.parent.joinpath(\"exported\", (\"sf_timevariantnetwork.geojson\")))\n\n# keep only selected columns fields\nsf_timevariantnetwork = sftimevariantnetwork[[\"id\", \"modes\",\"length\",\"lanes\",\"capacity\",\"geometry\",\n 'linkstats_freespeed','linkstats_volume', 'linkstats_traveltime', \n 'date_time', 'linkstats_freespeed_mph', 'linkstats_congspd_mph', 'linkstats_ratio', \"linkstats_vc_ratio\"]]\nsf_timevariantnetwork['date_time']=pd.to_datetime(sf_timevariantnetwork['date_time']).dt.strftime('%Y-%m-%dT%H:%M:%S')\nsf_timevariantnetwork[\"time\"] = pd.to_datetime(sf_timevariantnetwork[\"date_time\"]).dt.strftime('%Y-%m-%dT%H:%M:%S')\n\n\n# add more green shades for 85% --> 100%\n# green_shades = ['#008000', '#198c19', '#329932', '#4ca64c', '#66b266', '#7fbf7f', '#99cc99', '#b2d8b2', '#cce5cc', '#e5f2e5']\n# colors for congstd speed/freespeed ratio\ncolor_range_pct = [\"#ff0000\",\"#ff6666\",\"#ffb2b2\",\"#ffdb99\",\"#ffc966\", \"#ffa500\",'#e5f2e5','#cce5cc','#b2d8b2','#99cc99','#7fbf7f','#66b266','#4ca64c','#329932', '#198c19','#008000']\n# color_range_pct = [\"#ff0000\",\"#ff6666\",\"#ffb2b2\",\"#ffdb99\",\"#ffc966\", \"#ffa500\",\"#cce5cc\",\"#99cc99\",\"#66b266\",\"#008000\"]\nstep_pct = cmp.StepColormap(\n color_range_pct,\n vmin=0, vmax=1,\n index=[0,0.2,0.3,0.5,.6,0.7,0.80,0.85, 0.87,0.89,0.91,0.93,0.95,0.97,0.99,1.00], #for change in the colors, not used fr linear\n caption='% Speeds Difference' #Caption for Color scale or Legend\n)\n\n# colors for congstd speed/freespeed ratio\ncolor_range_pct_vc = ['#008000', '#329932', '#66b266', '#99cc99', '#cce5cc', '#e5f2e5', # green shade\n '#ffa500', \"#ffb732\",'#ffc966', '#ffdb99', \"#ffedcc\", # orange shade\n '#ffe5e5', '#ffcccc','#ffb2b2','#ff9999','#ff6666', '#ff3232', '#ff0000' ] # red shade\n# color_range_pct = [\"#ff0000\",\"#ff6666\",\"#ffb2b2\",\"#ffdb99\",\"#ffc966\", \"#ffa500\",\"#cce5cc\",\"#99cc99\",\"#66b266\",\"#008000\"]\nstep_pct_vc = cmp.StepColormap(\n color_range_pct_vc,\n vmin=0, vmax=1,\n index=[0,0.1,0.2,0.3,0.4,0.5,\n 0.55,0.6,0.65,0.7,0.75,\n 0.80,0.85,0.90,0.95,0.97,0.99,1.00], #for change in the colors, not used fr linear\n caption='Volume-to-Capacity ratio' #Caption for Color scale or Legend\n)\n\n# colors for congested speed and freespeed \ncolor_range = [\"#ff0000\",\"#ff6666\",\"#ffb2b2\",\"#ffa500\",\"#ffc966\",\"#ffdb99\", \"#cce5cc\",\"#99cc99\",\"#66b266\",\"#008000\"]\nstep = cmp.StepColormap(color_range,vmin=0, vmax=100,index=[0,5,10,15,25,35,45,55,65,100], #for change in the colors, not used fr linear\n caption=' Speeds (mph)' #Caption for Color scale or Legend\n )\n\n\ndef getColorMap_pct(x):\n return str(step_pct(x))\n\ndef getColorMap_pct_vc(x):\n return str(step_pct_vc(x))\n\ndef getColorMap(x):\n return str(step(x))\n\nsf_timevariantnetwork[\"fillColor_ratio\"] = sf_timevariantnetwork[\"linkstats_ratio\"].apply(getColorMap_pct)\nsf_timevariantnetwork[\"fillColor_vc_ratio\"] = sf_timevariantnetwork[\"linkstats_vc_ratio\"].apply(getColorMap_pct_vc)\nsf_timevariantnetwork[\"fillColor_freespeed_mph\"] = sf_timevariantnetwork[\"linkstats_freespeed_mph\"].apply(getColorMap)\nsf_timevariantnetwork[\"fillColor_congspd_mph\"] = sf_timevariantnetwork[\"linkstats_congspd_mph\"].apply(getColorMap)", "C:\\Users\\Transportlab\\anaconda3\\envs\\geo_env\\lib\\site-packages\\geopandas\\geodataframe.py:1351: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n super().__setitem__(key, value)\n" ], [ "def coords(geom):\n return list(geom.coords)\n\nsf_timevariantnetwork['points'] = sf_timevariantnetwork.apply(lambda row: coords(row.geometry), axis=1)\n\n# groupby and aggreage columns by segment_links\ndf1 = sf_timevariantnetwork.groupby('id').agg({'modes':'first',\n 'length':'first',\n 'lanes':list,\n 'capacity':list,\n 'geometry':'first',\n 'linkstats_freespeed':list,\n 'linkstats_volume':list,\n 'linkstats_traveltime':list,\n 'date_time':list,\n 'linkstats_freespeed_mph':list,\n 'linkstats_congspd_mph':list,\n 'linkstats_ratio':list,\n 'linkstats_vc_ratio':list,\n 'time':list,\n 'fillColor_ratio':list,\n 'linkstats_volume':list,\n 'linkstats_traveltime':list,\n 'fillColor_freespeed_mph':list,\n 'fillColor_congspd_mph':list,\n 'fillColor_vc_ratio':list,\n 'points':'first'}).reset_index()", "_____no_output_____" ], [ "# Create timemap for ratio_congestedspeed_freespeed\n\ndef coords(geom):\n return list(geom.coords)\n\nfeatures_ratio = [\n {\n 'type':'Feature',\n \"geometry\":{\n 'type': 'LineString',\n 'coordinates': coords(d.geometry),\n },\n 'properties': {\n 'times': d['time'],\n 'color': \"black\",\n 'colors':d[\"fillColor_ratio\"],\n \"weight\":0.6,\n \"fillOpacity\": 0.4,\n }\n }\n for _,d in df1.iterrows()\n]\n\nfrom jinja2 import Template\n_template = Template(\"\"\"\n {% macro script(this, kwargs) %} \n \n L.Control.TimeDimensionCustom = L.Control.TimeDimension.extend({\n _getDisplayDateFormat: function(date){\n var newdate = new moment(date);\n console.log(newdate)\n return newdate.format(\"{{this.date_options}}\");\n }\n });\n {{this._parent.get_name()}}.timeDimension = L.timeDimension(\n {\n period: {{ this.period|tojson }},\n }\n );\n var timeDimensionControl = new L.Control.TimeDimensionCustom(\n {{ this.options|tojson }}\n );\n {{this._parent.get_name()}}.addControl(this.timeDimensionControl);\n var geoJsonLayer = L.geoJson({{this.data}}, {\n pointToLayer: function (feature, latLng) {\n if (feature.properties.icon == 'marker') {\n if(feature.properties.iconstyle){\n return new L.Marker(latLng, {\n icon: L.icon(feature.properties.iconstyle)});\n }\n //else\n return new L.Marker(latLng);\n }\n if (feature.properties.icon == 'circle') {\n if (feature.properties.iconstyle) {\n return new L.circleMarker(latLng, feature.properties.iconstyle)\n };\n //else\n return new L.circleMarker(latLng);\n }\n //else\n return new L.Marker(latLng);\n },\n style: function(feature) {\n lastIdx=feature.properties.colors.length-1\n currIdx=feature.properties.colors.indexOf(feature.properties.color);\n if(currIdx==lastIdx){\n feature.properties.color = feature.properties.colors[currIdx+1] \n }\n else{\n feature.properties.color =feature.properties.colors[currIdx+1] \n }\n return {color: feature.properties.color}\n },\n onEachFeature: function(feature, layer) {\n if (feature.properties.popup) {\n layer.bindPopup(feature.properties.popup);\n }\n }\n })\n var {{this.get_name()}} = L.timeDimension.layer.geoJson(\n geoJsonLayer,\n {\n updateTimeDimension: true,\n addlastPoint: {{ this.add_last_point|tojson }},\n duration: {{ this.duration }},\n }\n ).addTo({{this._parent.get_name()}});\n {% endmacro %}\n \"\"\")\n\n\nimport folium\nfrom folium.plugins import TimestampedGeoJson\n\nm = folium.Map(location=[37.760015, -122.447110], zoom_start=13, tiles=\"cartodbpositron\")\n\nt=TimestampedGeoJson({\n 'type': 'FeatureCollection',\n 'features': features_ratio,\n}, transition_time=1500,loop=True,period='PT1H', add_last_point=False,auto_play=True)\nt._template=_template\nt.add_to(m)\nstep_pct.add_to(m)\n\n# Add title\nmap_title = \"Ratio between Congested Speed (mph) and Free Speed (mph)\"\ntitle_html = '''\n <h3 align=\"center\" style=\"font-size:16px\"><b>{}</b></h3>\n '''.format(map_title) \nm.get_root().html.add_child(folium.Element(title_html))\n\nfile_name = BASE_DIR.parent.joinpath(\"exported\", (\"linkstat_ratio_timemap.html\"))\nm.save(str(file_name))\n# m", "_____no_output_____" ], [ "# Create timemap for v/c ratio\n\ndef coords(geom):\n return list(geom.coords)\n\nfeatures_ratio = [\n {\n 'type':'Feature',\n \"geometry\":{\n 'type': 'LineString',\n 'coordinates': coords(d.geometry),\n },\n 'properties': {\n 'times': d['time'],\n 'color': \"black\",\n 'colors':d[\"fillColor_vc_ratio\"],\n \"weight\":0.6,\n \"fillOpacity\": 0.4,\n }\n }\n for _,d in df1.iterrows()\n]\n\nfrom jinja2 import Template\n_template = Template(\"\"\"\n {% macro script(this, kwargs) %} \n \n L.Control.TimeDimensionCustom = L.Control.TimeDimension.extend({\n _getDisplayDateFormat: function(date){\n var newdate = new moment(date);\n console.log(newdate)\n return newdate.format(\"{{this.date_options}}\");\n }\n });\n {{this._parent.get_name()}}.timeDimension = L.timeDimension(\n {\n period: {{ this.period|tojson }},\n }\n );\n var timeDimensionControl = new L.Control.TimeDimensionCustom(\n {{ this.options|tojson }}\n );\n {{this._parent.get_name()}}.addControl(this.timeDimensionControl);\n var geoJsonLayer = L.geoJson({{this.data}}, {\n pointToLayer: function (feature, latLng) {\n if (feature.properties.icon == 'marker') {\n if(feature.properties.iconstyle){\n return new L.Marker(latLng, {\n icon: L.icon(feature.properties.iconstyle)});\n }\n //else\n return new L.Marker(latLng);\n }\n if (feature.properties.icon == 'circle') {\n if (feature.properties.iconstyle) {\n return new L.circleMarker(latLng, feature.properties.iconstyle)\n };\n //else\n return new L.circleMarker(latLng);\n }\n //else\n return new L.Marker(latLng);\n },\n style: function(feature) {\n lastIdx=feature.properties.colors.length-1\n currIdx=feature.properties.colors.indexOf(feature.properties.color);\n if(currIdx==lastIdx){\n feature.properties.color = feature.properties.colors[currIdx+1] \n }\n else{\n feature.properties.color =feature.properties.colors[currIdx+1] \n }\n return {color: feature.properties.color}\n },\n onEachFeature: function(feature, layer) {\n if (feature.properties.popup) {\n layer.bindPopup(feature.properties.popup);\n }\n }\n })\n var {{this.get_name()}} = L.timeDimension.layer.geoJson(\n geoJsonLayer,\n {\n updateTimeDimension: true,\n addlastPoint: {{ this.add_last_point|tojson }},\n duration: {{ this.duration }},\n }\n ).addTo({{this._parent.get_name()}});\n {% endmacro %}\n \"\"\")\n\n\nimport folium\nfrom folium.plugins import TimestampedGeoJson\n\nm = folium.Map(location=[37.760015, -122.447110], zoom_start=13, tiles=\"cartodbpositron\")\n\nt=TimestampedGeoJson({\n 'type': 'FeatureCollection',\n 'features': features_ratio,\n}, transition_time=1500,loop=True,period='PT1H', add_last_point=False,auto_play=True)\nt._template=_template\nt.add_to(m)\nstep_pct_vc.add_to(m)\n\n# Add title\nmap_title = \"Volume-to-Capacity Ratio\"\ntitle_html = '''\n <h3 align=\"center\" style=\"font-size:16px\"><b>{}</b></h3>\n '''.format(map_title) \nm.get_root().html.add_child(folium.Element(title_html))\n\nfile_name = BASE_DIR.parent.joinpath(\"exported\",\"extended_run\",(\"linkst_vc_ratio_timemap.html\"))\nm.save(str(file_name))\n# m", "_____no_output_____" ], [ "# Create timemap for freespeed (mph)\n\ndef coords(geom):\n return list(geom.coords)\n\nfeatures_freespeed = [\n {\n 'type':'Feature',\n \"geometry\":{\n 'type': 'LineString',\n 'coordinates': coords(d.geometry),\n },\n 'properties': {\n 'times': d['time'],\n 'color': \"black\",\n 'colors':d[\"fillColor_freespeed_mph\"],\n 'weight':0.6,\n \"fillOpacity\": 0.4, \n }\n }\n for _,d in df1.iterrows()\n]\n\nfrom jinja2 import Template\n_template = Template(\"\"\"\n {% macro script(this, kwargs) %}\n L.Control.TimeDimensionCustom = L.Control.TimeDimension.extend({\n _getDisplayDateFormat: function(date){\n var newdate = new moment(date);\n console.log(newdate)\n return newdate.format(\"{{this.date_options}}\");\n }\n });\n {{this._parent.get_name()}}.timeDimension = L.timeDimension(\n {\n period: {{ this.period|tojson }},\n }\n );\n var timeDimensionControl = new L.Control.TimeDimensionCustom(\n {{ this.options|tojson }}\n );\n {{this._parent.get_name()}}.addControl(this.timeDimensionControl);\n var geoJsonLayer = L.geoJson({{this.data}}, {\n pointToLayer: function (feature, latLng) {\n if (feature.properties.icon == 'marker') {\n if(feature.properties.iconstyle){\n return new L.Marker(latLng, {\n icon: L.icon(feature.properties.iconstyle)});\n }\n //else\n return new L.Marker(latLng);\n }\n if (feature.properties.icon == 'circle') {\n if (feature.properties.iconstyle) {\n return new L.circleMarker(latLng, feature.properties.iconstyle)\n };\n //else\n return new L.circleMarker(latLng);\n }\n //else\n return new L.Marker(latLng);\n },\n style: function(feature) {\n lastIdx=feature.properties.colors.length-1\n currIdx=feature.properties.colors.indexOf(feature.properties.color);\n if(currIdx==lastIdx){\n feature.properties.color = feature.properties.colors[currIdx+1] \n }\n else{\n feature.properties.color =feature.properties.colors[currIdx+1] \n }\n return {color: feature.properties.color}\n },\n onEachFeature: function(feature, layer) {\n if (feature.properties.popup) {\n layer.bindPopup(feature.properties.popup);\n }\n }\n })\n var {{this.get_name()}} = L.timeDimension.layer.geoJson(\n geoJsonLayer,\n {\n updateTimeDimension: true,\n addlastPoint: {{ this.add_last_point|tojson }},\n duration: {{ this.duration }},\n }\n ).addTo({{this._parent.get_name()}});\n {% endmacro %}\n \"\"\")\n\n\nimport folium\nfrom folium.plugins import TimestampedGeoJson\n\nm = folium.Map(location=[37.760015, -122.447110], zoom_start=13, tiles=\"cartodbpositron\")\n# Add title\nmap_title = \"Free Speed (mph)\"\ntitle_html = '''\n <h3 align=\"center\" style=\"font-size:16px\"><b>{}</b></h3>\n '''.format(map_title) \nm.get_root().html.add_child(folium.Element(title_html))\n\nt=TimestampedGeoJson({\n 'type': 'FeatureCollection',\n 'features': features_freespeed,\n}, transition_time=1500,loop=True,period='PT1H', add_last_point=False,auto_play=True)\nt._template=_template\nt.add_to(m)\nstep.add_to(m)\n\nfile_name = BASE_DIR.parent.joinpath(\"exported\", (\"linkstat_freespeed_timemap.html\"))\nm.save(str(file_name))\n# m", "_____no_output_____" ], [ "# Create timemap for congested speed (mph)\n\ndef coords(geom):\n return list(geom.coords)\n\nfeatures_congestedspeed = [\n {\n 'type':'Feature',\n \"geometry\":{\n 'type': 'LineString',\n 'coordinates': coords(d.geometry),\n },\n 'properties': {\n 'times': d['time'],\n 'color': \"black\",\n 'colors':d[\"fillColor_congspd_mph\"],\n 'weight':0.6,\n \"fillOpacity\": 0.4,\n \n }\n }\n for _,d in df1.iterrows()\n]\n\nfrom jinja2 import Template\n_template = Template(\"\"\"\n {% macro script(this, kwargs) %}\n L.Control.TimeDimensionCustom = L.Control.TimeDimension.extend({\n _getDisplayDateFormat: function(date){\n var newdate = new moment(date);\n console.log(newdate)\n return newdate.format(\"{{this.date_options}}\");\n }\n });\n {{this._parent.get_name()}}.timeDimension = L.timeDimension(\n {\n period: {{ this.period|tojson }},\n }\n );\n var timeDimensionControl = new L.Control.TimeDimensionCustom(\n {{ this.options|tojson }}\n );\n {{this._parent.get_name()}}.addControl(this.timeDimensionControl);\n var geoJsonLayer = L.geoJson({{this.data}}, {\n pointToLayer: function (feature, latLng) {\n if (feature.properties.icon == 'marker') {\n if(feature.properties.iconstyle){\n return new L.Marker(latLng, {\n icon: L.icon(feature.properties.iconstyle)});\n }\n //else\n return new L.Marker(latLng);\n }\n if (feature.properties.icon == 'circle') {\n if (feature.properties.iconstyle) {\n return new L.circleMarker(latLng, feature.properties.iconstyle)\n };\n //else\n return new L.circleMarker(latLng);\n }\n //else\n return new L.Marker(latLng);\n },\n style: function(feature) {\n lastIdx=feature.properties.colors.length-1\n currIdx=feature.properties.colors.indexOf(feature.properties.color);\n if(currIdx==lastIdx){\n feature.properties.color = feature.properties.colors[currIdx+1] \n }\n else{\n feature.properties.color =feature.properties.colors[currIdx+1] \n }\n return {color: feature.properties.color}\n },\n onEachFeature: function(feature, layer) {\n if (feature.properties.popup) {\n layer.bindPopup(feature.properties.popup);\n }\n }\n })\n var {{this.get_name()}} = L.timeDimension.layer.geoJson(\n geoJsonLayer,\n {\n updateTimeDimension: true,\n addlastPoint: {{ this.add_last_point|tojson }},\n duration: {{ this.duration }},\n }\n ).addTo({{this._parent.get_name()}});\n {% endmacro %}\n \"\"\")\n\n\nimport folium\nfrom folium.plugins import TimestampedGeoJson\n\nm = folium.Map(location=[37.760015, -122.447110], zoom_start=13, tiles=\"cartodbpositron\")\n\n\nt=TimestampedGeoJson({\n 'type': 'FeatureCollection',\n 'features': features_congestedspeed,\n}, transition_time=1500,loop=True,period='PT1H', add_last_point=False,auto_play=True)\nt._template=_template\nt.add_to(m)\nstep.add_to(m)\n\n# Add title\nmap_title = \"Congested Speed (mph)\"\ntitle_html = '''\n <h3 align=\"center\" style=\"font-size:16px\"><b>{}</b></h3>\n '''.format(map_title) \nm.get_root().html.add_child(folium.Element(title_html))\n\n\nfile_name = BASE_DIR.parent.joinpath(\"exported\", (\"linkstat_congestedspeed_timemap.html\"))\nm.save(str(file_name))\n# m", "_____no_output_____" ], [ "# get map for each different time", "_____no_output_____" ], [ "# static maps for congestedspeed/freespeed\n\ndef get_dataframe(_time):\n # linkstats file\n linkstats = pd.read_csv(\"../SF_all_trips/sf-tscore-all-trips-20PCsample-updatedRideHailFleet-updatedParking__etg/ITERS/it.4/4.linkstats.csv.gz\", compression=\"gzip\", low_memory=False)\n# unmodified_linkstats = pd.read_csv(BASE_DIR.parent.joinpath(\"runs\", \"sf-tscore-int-int-trips-model-network-events-20PC-sample-bpr-func__tlm\",\"ITERS\",\"it.30\", \"30.linkstats_unmodified.csv.gz\"),compression=\"gzip\", low_memory=False)\n time = int(_time)\n linkstats = linkstats[linkstats[\"hour\"]==(time)].copy()\n linkstats=linkstats.add_prefix(\"linkstats_\")\n linkstats.rename(columns={('linkstats_link'): 'id'}, inplace=True)\n linkstats[\"id\"] = linkstats[\"id\"].astype('string')\n \n return linkstats\n\ndef highlight_function(feature):\n return {\"fillColor\": \"#ffff00\", \"color\": \"#ffff00\", \"weight\": 5,\"fillOpacity\": 0.40 }\n\n\ncolor_range_pct = [\"#ff0000\",\"#ff6666\",\"#ffb2b2\",\"#ffdb99\",\"#ffc966\", \"#ffa500\",'#e5f2e5','#cce5cc','#b2d8b2','#99cc99','#7fbf7f','#66b266','#4ca64c','#329932', '#198c19','#008000']\n\nstep_pct = cmp.StepColormap(\n color_range_pct,\n vmin=0, vmax=1,\n index=[0,0.2,0.3,0.5,.6,0.7,0.80,0.85, 0.87,0.89,0.91,0.93,0.95,0.97,0.99,1.00], #for change in the colors, not used fr linear\n caption='% Speeds Difference' #Caption for Color scale or Legend\n)\n\n# read the road network\nsf_roadnetwork = gpd.read_file(BASE_DIR.parent.joinpath(\"Network\", \"sfNetwork.geojsonl\"))\nsf_roadnetwork = sf_roadnetwork[[\"id\",\"modes\",\"length\",\"lanes\",\"from\",\"to\",\"capacity\",\"geometry\"]]\n\nfor time_hour in tqdm(range(0,30)):\n # set the map\n pct_m = folium.Map([37.760015, -122.447110], zoom_start=13, tiles=\"cartodbpositron\")\n # get the hour and filter the linkstat file\n linkstats = get_dataframe(str(time_hour))\n # merge with featureclass of SF data\n comparision_network = sf_roadnetwork.merge(linkstats,on=\"id\")\n \n # calculate the freespeed (mph), congested speed (mph), ratio (congestedsped/freespeed)\n # linkstats\n comparision_network[\"linkstats_freespeed_mph\"] = comparision_network[\"linkstats_freespeed\"]*2.23694\n comparision_network[\"linkstats_congspd_mph\"] = (comparision_network[\"linkstats_length\"]/comparision_network[\"linkstats_traveltime\"])*2.23694\n comparision_network[\"linkstats_ratio\"] = comparision_network[\"linkstats_congspd_mph\"] / comparision_network[\"linkstats_freespeed_mph\"]\n \n time_stamp = \"\"\n # folium\n if time_hour<30: \n time_stamp = f'{time_hour:02d}'\n layer_name = str(time_stamp) \n \n \n# layer_name=str(str(time_hour) + (' am' if time_hour < 12 else ' pm'))\n ratio_feature_group = folium.FeatureGroup(name=layer_name)\n \n pct_feature_group = folium.GeoJson(comparision_network, \n name = (\"Hour - \" + layer_name),\n style_function=lambda x: {\n \"fillColor\": step_pct(x[\"properties\"][\"linkstats_ratio\"]),\n \"color\": step_pct(x[\"properties\"][\"linkstats_ratio\"]),\n \"fillOpacity\": 0.2,\n \"weight\":1,\n },\n tooltip=folium.GeoJsonTooltip(fields=[\"id\",\"length\", \"linkstats_freespeed_mph\", \"linkstats_traveltime\",\"linkstats_congspd_mph\"], \n aliases=[\"Link ID\", \"Segment Length (m)\", \"Freespeed (mph)\", \"Travel time (sec)\", \"Congested Speed (mph)\"], localize=True),\n popup = folium.GeoJsonPopup(fields=[\"id\",\"length\", \"linkstats_freespeed_mph\", \"linkstats_traveltime\",\"linkstats_congspd_mph\"], \n aliases=[\"Link ID\", \"Segment Length (m)\", \"Freespeed (mph)\", \"Travel time (sec)\", \"Congested Speed (mph)\"], localize=True),\n highlight_function=highlight_function,\n zoom_on_click=True\n ).add_to(ratio_feature_group)\n # Add search functionality to the map\n search_link = Search(layer=pct_feature_group, geom_type=\"LineString\", placeholders = \"Search for Link ID\", \n collapsed=\"False\", search_label = 'id', search_zoom = 17, position='topleft',\n ).add_to(pct_m)\n \n ratio_feature_group.add_to(pct_m)\n folium.LayerControl().add_to(pct_m)\n \n map_title = \"Ratio between Congested Speed and Free Speed\"\n title_html = '''<h3 align=\"center\" style=\"font-size:16px\"><b>{}</b></h3>'''.format(map_title)\n pct_m.get_root().html.add_child(folium.Element(title_html))\n # save the file\n file_name = BASE_DIR.parent.joinpath(\"exported\", (\"linkstat_ratio_timemap_{}.html\").format(time_stamp))\n pct_m.save(str(file_name))", "100%|██████████████████████████████████████████████████████████████████████████████████| 30/30 [32:58<00:00, 65.95s/it]\n" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "<a href=\"https://colab.research.google.com/github/malcolmrite-dsi/RockVideoClassifier/blob/main/RocksResnetTrainer.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>", "_____no_output_____" ] ], [ [ "from google.colab import drive\ndrive.mount('/content/drive')", "Mounted at /content/drive\n" ], [ "import tensorflow as tf\nimport numpy as np\nfrom tensorflow import keras\nfrom tensorflow.keras.models import Model\nfrom tensorflow.keras.layers import Dense\nfrom tensorflow.keras.layers import Flatten, Dropout\nfrom tensorflow.keras.applications.resnet import preprocess_input\nfrom tensorflow.keras.applications import xception\nimport pandas as pd\nimport PIL\nimport matplotlib.pyplot as plt\nimport os\nimport shutil", "_____no_output_____" ] ], [ [ "# Training", "_____no_output_____" ] ], [ [ "train_datagen = keras.preprocessing.image.ImageDataGenerator(validation_split=0.2, preprocessing_function=preprocess_input)\n\ntrain_generator = train_datagen.flow_from_directory(\n '/content/drive/My Drive/Module 2 shared folder/samples',\n subset=\"training\",\n seed=3,\n target_size=(64, 64),\n batch_size=64,\n class_mode='categorical')\n\nval_generator = train_datagen.flow_from_directory( '/content/drive/My Drive/Module 2 shared folder/samples', \n subset=\"validation\",\n seed=3,\n target_size=(64, 64),\n batch_size=64,\n class_mode='categorical')", "Found 3132 images belonging to 5 classes.\nFound 781 images belonging to 5 classes.\n" ], [ "train_datagen = keras.preprocessing.image.ImageDataGenerator(preprocessing_function=preprocess_input)\n\ntrain_generator = train_datagen.flow_from_directory(\n '/content/drive/My Drive/Module 2 shared folder/samples',\n subset=\"training\",\n seed=3,\n target_size=(64, 64),\n batch_size=64,\n class_mode='categorical')", "Found 3913 images belonging to 5 classes.\n" ], [ "resnet = keras.applications.ResNet50(include_top=False, pooling=\"max\", input_shape=(64,64,3))", "_____no_output_____" ], [ "# mark loaded layers as not trainable\nfor layer in resnet.layers:\n layer.trainable = False \n", "_____no_output_____" ], [ "data_augmentation = tf.keras.Sequential([\n keras.layers.experimental.preprocessing.RandomFlip(\"horizontal_and_vertical\"),\n #keras.layers.experimental.preprocessing.RandomRotation(0.2),\n])", "_____no_output_____" ], [ "# mark loaded layers as not trainable\n#for layer in resnet.layers:\n\t#layer.trainable = False \n\nflat = Flatten()(resnet.layers[-1].output)\ndense = Dense(1024, activation='relu')(flat)\noutput = Dense(5, activation='softmax')(dense)\nmodel = Model(inputs=resnet.inputs, outputs=output)", "_____no_output_____" ], [ "model.summary()", "_____no_output_____" ], [ "model.compile(loss=\"categorical_crossentropy\", optimizer=keras.optimizers.Adam(), metrics=[\"categorical_accuracy\"])\ncheckpoint_best = keras.callbacks.ModelCheckpoint(\"/content/drive/My Drive/model_best.h5\", \n monitor='loss', verbose=0, save_best_only=True, save_weights_only=False, save_freq='epoch')\ncheckpoint = keras.callbacks.ModelCheckpoint(\"/content/drive/My Drive/model_last.h5\", \n verbose=0, save_best_only=False, save_weights_only=False, save_freq='epoch')\nmodel.fit(\n train_generator,\n epochs = 5,\n validation_data=val_generator,\n callbacks=[checkpoint_best]\n)", "Epoch 1/5\n49/49 [==============================] - 71s 1s/step - loss: 1.1823 - categorical_accuracy: 0.8065 - val_loss: 0.7923 - val_categorical_accuracy: 0.8502\nEpoch 2/5\n49/49 [==============================] - 72s 1s/step - loss: 0.3338 - categorical_accuracy: 0.9013 - val_loss: 0.5771 - val_categorical_accuracy: 0.8668\nEpoch 3/5\n49/49 [==============================] - 72s 1s/step - loss: 0.1848 - categorical_accuracy: 0.9377 - val_loss: 0.5730 - val_categorical_accuracy: 0.8758\nEpoch 4/5\n49/49 [==============================] - 75s 2s/step - loss: 0.1533 - categorical_accuracy: 0.9496 - val_loss: 0.7109 - val_categorical_accuracy: 0.8630\nEpoch 5/5\n49/49 [==============================] - 73s 1s/step - loss: 0.1125 - categorical_accuracy: 0.9642 - val_loss: 0.7830 - val_categorical_accuracy: 0.8643\n" ], [ "model.evaluate(val_generator)", "13/13 [==============================] - 42s 3s/step - loss: 0.5171 - categorical_accuracy: 0.8912\n" ], [ "model.fit(\n train_generator,\n initial_epoch=10,\n epochs = 20,\n validation_data=val_generator, callbacks=[checkpoint, checkpoint_best]\n)", "Epoch 11/20\n62/62 [==============================] - 87s 1s/step - loss: 0.3099 - categorical_accuracy: 0.9213 - val_loss: 0.3068 - val_categorical_accuracy: 0.9117\nEpoch 12/20\n62/62 [==============================] - 86s 1s/step - loss: 0.1898 - categorical_accuracy: 0.9397 - val_loss: 0.1291 - val_categorical_accuracy: 0.9565\nEpoch 13/20\n62/62 [==============================] - 88s 1s/step - loss: 0.1054 - categorical_accuracy: 0.9635 - val_loss: 0.1378 - val_categorical_accuracy: 0.9693\nEpoch 14/20\n62/62 [==============================] - 88s 1s/step - loss: 0.0665 - categorical_accuracy: 0.9793 - val_loss: 0.0363 - val_categorical_accuracy: 0.9885\nEpoch 15/20\n62/62 [==============================] - 89s 1s/step - loss: 0.0444 - categorical_accuracy: 0.9875 - val_loss: 0.0727 - val_categorical_accuracy: 0.9706\nEpoch 16/20\n62/62 [==============================] - 87s 1s/step - loss: 0.0873 - categorical_accuracy: 0.9683 - val_loss: 0.0584 - val_categorical_accuracy: 0.9821\nEpoch 17/20\n62/62 [==============================] - 90s 1s/step - loss: 0.0384 - categorical_accuracy: 0.9885 - val_loss: 0.0240 - val_categorical_accuracy: 0.9923\nEpoch 18/20\n62/62 [==============================] - 87s 1s/step - loss: 0.0466 - categorical_accuracy: 0.9842 - val_loss: 0.0385 - val_categorical_accuracy: 0.9910\nEpoch 19/20\n62/62 [==============================] - 88s 1s/step - loss: 0.0379 - categorical_accuracy: 0.9880 - val_loss: 0.0192 - val_categorical_accuracy: 0.9949\nEpoch 20/20\n62/62 [==============================] - 89s 1s/step - loss: 0.0262 - categorical_accuracy: 0.9928 - val_loss: 0.0225 - val_categorical_accuracy: 0.9910\n" ], [ "model.save(\"/content/drive/My Drive/model_best_64.h5\")", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "%run scripts/train_models/launch_experiments.py --experiment_id=test --random", "main folder:/home/jarobyte/scratch/guemes/icdar/es/\noutput folder:baseline\n\nModel: Transformer\nTokens in the input vocabulary: 182\nTokens in the output vocabulary: 182\nMax sequence length: 110\nEmbedding dimension: 256\nFeedforward dimension: 1024\nEncoder layers: 2\nDecoder layers: 2\nAttention heads: 8\nActivation: relu\nDropout: 0.4\nTrainable parameters: 3,855,542\n\nTraining started\nEpochs: 10\nLearning rate: 0.0001\nWeight decay: 0.001\nEpoch | Train | Development | Minutes\n | Loss | Error Rate | Loss | Error Rate |\n---------------------------------------------------------------\n 1 | 4.4012 | 83.244 | 4.2320 | 99.960 | 0.0\n 2 | 3.5844 | 77.707 | 3.8039 | 99.089 | 0.0\n 3 | 3.3182 | 77.088 | 3.6094 | 92.277 | 0.0\n 4 | 3.1461 | 75.265 | 3.4594 | 85.485 | 0.0\n 5 | 3.0011 | 72.756 | 3.3018 | 80.733 | 0.0\n 6 | 2.8745 | 70.236 | 3.1664 | 78.208 | 0.1\n 7 | 2.7630 | 68.191 | 3.0840 | 78.000 | 0.1\n 8 | 2.6652 | 66.625 | 3.0250 | 77.842 | 0.1\n 9 | 2.5855 | 65.606 | 2.9725 | 77.554 | 0.1\n 10 | 2.5184 | 64.743 | 2.9266 | 77.337 | 0.1\n\nEvaluating model..\ndisjoint window...\ngreedy_search...\nbeam_search...\nsliding, greedy...\nuniform...\ntriangle...\nbell...\nsliding, beam...\nuniform...\ntriangle...\nbell...\n experiment_id improvement window decoding window_size \\\n0 test -70.150204 disjoint greedy 10 \n1 test -67.381259 disjoint greedy 5 \n2 test -79.538192 disjoint beam 10 \n3 test -78.658891 disjoint beam 5 \n4 test -72.055369 sliding greedy 5 \n5 test -71.941569 sliding greedy 5 \n6 test -71.941569 sliding greedy 5 \n7 test -78.131311 sliding beam 5 \n8 test -78.131311 sliding beam 5 \n9 test -78.131311 sliding beam 5 \n\n inference_seconds cer_before cer_after weighting \n0 0.647468 43.407167 73.857383 NaN \n1 0.584162 43.407167 72.655462 NaN \n2 4.998648 43.407167 77.932442 NaN \n3 4.504213 43.407167 77.550763 NaN \n4 4.149696 43.407167 74.684361 uniform \n5 3.930020 43.407167 74.634964 triangle \n6 3.926667 43.407167 74.634964 bell \n7 23.705193 43.407167 77.321755 uniform \n8 23.494794 43.407167 77.321755 triangle \n9 23.261851 43.407167 77.321755 bell \n" ] ] ]

[ "code" ]

[ [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "<img src='./img/EU-Copernicus-EUM_3Logos.png' alt='Logo EU Copernicus EUMETSAT' align='right' width='40%'></img>", "_____no_output_____" ], [ "<br>", "_____no_output_____" ], [ "<a href=\"./00_index.ipynb\"><< Index</a><span style=\"float:right;\"><a href=\"./02_AC_SAF_GOME-2_L2_produce_gridded_dataset_L3.ipynb\">02 - AC SAF GOME-2 - Produce gridded dataset (L3)>></a> ", "_____no_output_____" ], [ "<br>", "_____no_output_____" ], [ "# Optional: Introduction to Python and Project Jupyter", "_____no_output_____" ], [ "## Project Jupyter", "_____no_output_____" ], [ "<div class=\"alert alert-block alert-success\" align=\"center\">\n<b><i>\"Project Jupyter exists to develop open-source software, open-standards, and services for interactive computing across dozens of programming languages.\"</i></b>\n</div>", "_____no_output_____" ], [ "<br>", "_____no_output_____" ], [ "Project Jupyter offers different tools to facilitate interactive computing, either with a web-based application (`Jupyter Notebooks`), an interactive development environment (`JupyterLab`) or via a `JupyterHub` that brings interactive computing to groups of users.\n\n<br>", "_____no_output_____" ], [ "<center><img src='./img/jupyter_environment.png' alt='Logo Jupyter environment' width='60%'></img></center>", "_____no_output_____" ], [ "* **Jupyter Notebook** is an open-source web application that allows you to create and share documents that contain live code, equations, visualizations and narrative text.\n\n* **JupyterLab 1.0: Jupyter’s Next-Generation Notebook Interface** <br> JupyterLab is a web-based interactive development environment for Jupyter notebooks, code, and data.\n\n* **JupyterHub** <br>JupyterHub brings the power of notebooks to groups of users. It gives users access to computational environments and resources without burdening the users with installation and maintenance tasks. <br> Users - including students, researchers, and data scientists - can get their work done in their own workspaces on shared resources which can be managed efficiently by system administrators.", "_____no_output_____" ], [ "<hr>", "_____no_output_____" ], [ "## Why Jupyter Notebooks?", "_____no_output_____" ], [ "* Started with Python support, now **support of over 40 programming languages, including Python, R, Julia, ...**\n* Notebooks can **easily be shared via GitHub, NBViewer, etc.**\n* **Code, data and visualizations are combined in one place**\n* A great tool for **teaching**\n* **JupyterHub allows you to access an environment ready to code**", "_____no_output_____" ], [ "## Installation", "_____no_output_____" ], [ "### Installing Jupyter using Anaconda", "_____no_output_____" ], [ "Anaconda comes with the Jupyter Notebook installed. You just have to download Anaconda and following the installation instructions. Once installed, the jupyter notebook can be started with:", "_____no_output_____" ] ], [ [ "jupyter notebook", "_____no_output_____" ] ], [ [ "### Installing Jupyter with pip", "_____no_output_____" ], [ "Experienced Python users may want to install Jupyter using Python's package manager `pip`.\n\nWith `Python3` you do:", "_____no_output_____" ] ], [ [ "python3 -m pip install --upgrade pip\npython3 -m pip install jupyter", "_____no_output_____" ] ], [ [ "In order to run the notebook, you run the same command as with Anaconda at the Terminal :", "_____no_output_____" ] ], [ [ "jupyter notebook", "_____no_output_____" ] ], [ [ "## Jupyter notebooks UI", "_____no_output_____" ], [ "* Notebook dashboard\n * Create new notebook\n* Notebook editor (UI)\n * Menu\n * Toolbar\n * Notebook area and cells\n* Cell types\n * Code\n * Markdown\n* Edit (green) vs. Command mode (blue)", "_____no_output_____" ], [ "<br>\n\n\n<div style='text-align:center;'>\n<figure><img src='./img/notebook_ui.png' width='100%'/>\n <figcaption><i>Notebook editor User Interface (UI)</i></figcaption>\n</figure>\n</div>", "_____no_output_____" ], [ "## Shortcuts", "_____no_output_____" ], [ "Get an overview of the shortcuts by hitting `H` or go to `Help/Keyboard shortcuts`", "_____no_output_____" ], [ "#### Most useful shortcuts", "_____no_output_____" ], [ "* `Esc` - switch to command mode\n* `B` - insert below\n* `A` - insert above\n* `M` - Change current cell to Markdown\n* `Y` - Change current cell to code\n* `DD` - Delete cell\n* `Enter` - go back to edit mode\n* `Esc + F` - Find and replace on your code\n* `Shift + Down / Upwards` - Select multiple cells\n* `Shift + M` - Merge multiple cells", "_____no_output_____" ], [ "## Cell magics", "_____no_output_____" ], [ "Magic commands can make your life a lot easier, as you only have one command instead of an entire function or multiple lines of code.<br>\n> Go to an [extensive overview of magic commands]()", "_____no_output_____" ], [ "### Some of the handy ones", "_____no_output_____" ], [ "**Overview of available magic commands**", "_____no_output_____" ] ], [ [ "%lsmagic", "_____no_output_____" ] ], [ [ "**See and set environment variables**", "_____no_output_____" ] ], [ [ "%env", "_____no_output_____" ] ], [ [ "**Install and list libraries**", "_____no_output_____" ] ], [ [ "!pip install numpy\n!pip list | grep pandas", "_____no_output_____" ] ], [ [ "**Write cell content to a Python file**", "_____no_output_____" ] ], [ [ "%%writefile hello_world.py\n\nprint('Hello World')", "_____no_output_____" ] ], [ [ "**Load a Python file**", "_____no_output_____" ] ], [ [ "%pycat hello_world.py", "_____no_output_____" ] ], [ [ "**Get the time of cell execution**", "_____no_output_____" ] ], [ [ "%%time\n\ntmpList = []\nfor i in range(100):\n tmpList.append(i+i)\n\nprint(tmpList)", "_____no_output_____" ] ], [ [ "**Show matplotlib plots inline**", "_____no_output_____" ] ], [ [ "%matplotlib inline", "_____no_output_____" ] ], [ [ "<br>", "_____no_output_____" ], [ "## Sharing Jupyter Notebooks", "_____no_output_____" ], [ "### Sharing static Jupyter Notebooks", "_____no_output_____" ], [ "* [nbviewer](https://nbviewer.jupyter.org/) - A simple way to share Jupyter Notebooks. You can simply paste the GitHub location of your Jupyter notebook there and it is nicely rendered.\n* [GitHub](https://github.com/) - GitHub offers an internal rendering of Jupyter Notebooks. There are some limitations and time delays of the proper rendering. Thus, we would suggest to use `nbviewer` to share nicely rendered Jupyter Notebooks.", "_____no_output_____" ], [ "### Reproducible Jupyter Notebooks", "_____no_output_____" ], [ "<img src=\"./img/mybinder_logo.png\" align=\"left\" width=\"30%\"></img>\n[Binder](https://mybinder.org/) allows you to open notebooks hosted on a Git repo in an executable environment, making the code immediately reproducible by anyone, anywhere.\n\nBinder builds a Docker image of the repo where the notebooks are hosted.\n", "_____no_output_____" ], [ "<br>", "_____no_output_____" ], [ "## Ressources", "_____no_output_____" ], [ "* [Project Jupyter](https://jupyter.org/)\n* [JupyterHub](https://jupyterhub.readthedocs.io/en/stable/)\n* [JupyterLab](https://jupyterlab.readthedocs.io/en/stable/)\n* [nbviewer](https://nbviewer.jupyter.org/)\n* [Binder](https://mybinder.org/)", "_____no_output_____" ], [ "<br>", "_____no_output_____" ], [ "<a href=\"./00_index.ipynb\"><< Index</a><span style=\"float:right;\"><a href=\"./02_AC_SAF_GOME-2_L2_produce_gridded_dataset_L3.ipynb\">02 - AC SAF GOME-2 - Produce gridded dataset (L3)>></a> ", "_____no_output_____" ], [ "<hr>", "_____no_output_____" ], [ "<img src='./img/copernicus_logo.png' alt='Logo EU Copernicus' align='right' width='20%'><br><br><br>\n<br>\n<p style=\"text-align:right;\">This project is licensed under the <a href=\"./LICENSE\">MIT License</a> and is developed under a Copernicus contract.", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# reference https://socratic.org/questions/the-top-and-bottom-margins-of-a-poster-are-4-cm-and-the-side-margins-are-each-6-\nfrom IPython.display import Image\nfrom IPython.core.display import HTML \nfrom sympy import *; x,h,y,n,t = symbols(\"x h y n t\"); C, D = symbols(\"C D\", real=True)\nImage(url= \"https://i.imgur.com/LA701Wi.png\")", "_____no_output_____" ], [ "A,a,b = symbols(\"A a b\")\nA = 382 + 2*(a*8) + 2*(b*8)-4*(8*8)\nA", "_____no_output_____" ], [ "Eq(A,a*b)", "_____no_output_____" ], [ "pprint(solve(Eq(A,a*b),a)) # A(b)", "⎡2⋅(8⋅b + 63)⎤\n⎢────────────⎥\n⎣ b - 16 ⎦\n" ], [ "simplify((2*(8*b + 63)/(b - 16))*b) *a * b", "_____no_output_____" ], [ "simplify(diff((2*(8*b + 63)/(b - 16))*b))", "_____no_output_____" ], [ "print(simplify(diff((2*(8*b + 63)/(b - 16))*b)))", "16*(b**2 - 32*b - 126)/(b**2 - 32*b + 256)\n" ], [ "b**2 - 32*b - 126", "_____no_output_____" ], [ "solve(b**2 - 32*b - 126,b)", "_____no_output_____" ], [ "Image(url= \"https://i.imgur.com/3RToGO0.png\")", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "from utils import *", "_____no_output_____" ], [ "from azureml.core import Workspace\nfrom azureml.core.authentication import InteractiveLoginAuthentication\n\ninteractive_auth = InteractiveLoginAuthentication(tenant_id=\"c643d8c8-4abb-47ba-808c-bf73f60161f1\")\n\n# Configure workspace\nws = Workspace.from_config(auth=interactive_auth)", "_____no_output_____" ], [ "from azureml.core.model import InferenceConfig\n\nenv = get_env(name=\"sentiment-analysis\", packages=[\"pandas\", \"inference-schema\", \"tensorflow\", \"transformers\"])\n\ninference_config = InferenceConfig(\n environment=env,\n source_directory=\"code\",\n entry_script=\"sentiment_analysis_pandas.py\",\n)", "_____no_output_____" ], [ "from azureml.core.webservice import LocalWebservice\n\ndeployment_config = LocalWebservice.deploy_configuration(port=6789)", "_____no_output_____" ], [ "from azureml.core import Model\n\nservice = Model.deploy(\n ws,\n \"sentiment-analysis-pandas\",\n [],\n inference_config,\n deployment_config,\n overwrite=True,\n)\nservice.wait_for_deployment(show_output=True)", "_____no_output_____" ], [ "print(service.get_logs())", "_____no_output_____" ], [ "import requests\nimport json\n\nuri = service.scoring_uri\nrequests.get(\"http://localhost:6789\")\nheaders = {\"Content-Type\": \"application/json\"}\ndata = {\n \"Inputs\": {\n \"data\": [\n {\"query\": \"My ML skills are quite good.\"}, \n {\"query\": \"I didn't have good experience with ML.\"}\n ]\n }\n}\ndata = json.dumps(data)\nresponse = requests.post(uri, data=data, headers=headers)\nprint(response.text)", "_____no_output_____" ], [ "# service.delete()", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "Adding functionality so we can create dataframe from string representation of dict", "_____no_output_____" ] ], [ [ "import ast\ndef str_to_dict(string):\n return ast.literal_eval(string) ", "_____no_output_____" ], [ "import pandas as pd", "_____no_output_____" ], [ "class MySubClass(pd.DataFrame):\n def from_str(self, string):\n df_obj = super().from_dict(str_to_dict(string))\n df_obj.my_string_attribute = string\n return df_obj", "_____no_output_____" ], [ "data = \"{'col_1' : ['a','b'], 'col2': [1, 2]}\"", "_____no_output_____" ], [ "obj = MySubClass().from_str(data)", "_____no_output_____" ], [ "type(obj)", "_____no_output_____" ], [ "obj", "_____no_output_____" ], [ "obj.my_string_attribute", "_____no_output_____" ], [ "sales = [{'account': 'Jones LLC', 'Jan': 150, 'Feb': 200, 'Mar': 140},\n {'account': 'Alpha Co', 'Jan': 200, 'Feb': 210, 'Mar': 215},\n {'account': 'Blue Inc', 'Jan': 50, 'Feb': 90, 'Mar': 95 }]\ndf = MySubClass(sales)", "_____no_output_____" ], [ "df", "_____no_output_____" ], [ "type(df)", "_____no_output_____" ] ] ]

[ "raw", "code" ]

[ [ "raw" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "<center>\n<a href=\"http://www.insa-toulouse.fr/\" ><img src=\"http://www.math.univ-toulouse.fr/~besse/Wikistat/Images/logo-insa.jpg\" style=\"float:left; max-width: 120px; display: inline\" alt=\"INSA\"/></a> \n\n<a href=\"http://wikistat.fr/\" ><img src=\"http://www.math.univ-toulouse.fr/~besse/Wikistat/Images/wikistat.jpg\" style=\"max-width: 250px; display: inline\" alt=\"Wikistat\"/></a>\n\n<a href=\"http://www.math.univ-toulouse.fr/\" ><img src=\"http://www.math.univ-toulouse.fr/~besse/Wikistat/Images/logo_imt.jpg\" style=\"float:right; max-width: 200px; display: inline\" alt=\"IMT\"/> </a>\n</center>", "_____no_output_____" ], [ "# [Ateliers: Technologies des grosses données](https://github.com/wikistat/Ateliers-Big-Data)", "_____no_output_____" ], [ "# Recommandation de Films par Filtrage Collaboratif: [NMF](http://wikistat.fr/pdf/st-m-explo-nmf.pdf) de la librairie [SparkML](https://spark.apache.org/docs/latest/ml-guide.html) de <a href=\"http://spark.apache.org/\"><img src=\"http://spark.apache.org/images/spark-logo-trademark.png\" style=\"max-width: 100px; display: inline\" alt=\"Spark\"/></a>", "_____no_output_____" ], [ "## 1. Introduction\n\nCe calepin traite d'un problème classique de recommandation par filtrage collaboratif en utilisant les ressources de la librairie [MLlib de Spark]([http://spark.apache.org/docs/latest/api/python/pyspark.mllib.html#pyspark.mllib.recommendation.ALS) avec l'API pyspark. Le problème général est décrit en [introduction](https://github.com/wikistat/Ateliers-Big-Data/tree/master/3-MovieLens) et dans une [vignette](http://wikistat.fr/pdf/st-m-datSc3-colFil.pdf) de [Wikistat](http://wikistat.fr/). Il est appliqué aux données publiques du site [GroupLens](http://grouplens.org/datasets/movielens/). L'objectif est de tester les méthodes et la procédure d'optimisation sur le plus petit jeu de données composé de 100k notes de 943 clients sur 1682 films où chaque client a au moins noté 20 films. Les jeux de données plus gros (1M, 10M, 20M notes) peuvent être utilisés pour \"passer à l'échelle volume\". \n\nCe calepin s'inspire des exemples de la [documentation](http://spark.apache.org/docs/latest/api/python/pyspark.mllib.html#pyspark.mllib.recommendation.ALS) et d'un [tutoriel](https://github.com/jadianes/spark-movie-lens/blob/master/notebooks/building-recommender.ipynb) de [Jose A. Dianes](https://www.codementor.io/jadianes). Le sujet a été traité lors d'un [Spark Summit](https://databricks-training.s3.amazonaws.com/movie-recommendation-with-mllib.html).\n\nL'objectif est d'utiliser ces seules données pour proposer des recommandations. Les données initiales sont sous la forme d'une matrice **très creuse** (*sparse*) contenant des notes ou évaluations. **Attention**, les \"0\" de la matrice ne sont pas des notes mais des *données manquantes*, le film n'a pas encore été vu ou évalué. \n\nUn algorithme satisfaisant à l'objectif de *complétion de grande matrice creuse*, et implémenté dans un logiciel libre d'accès est disponible dans la librairie [softImpute de R](https://cran.r-project.org/web/packages/softImpute/index.html). SOn utilisaiton est décrite dans un autre [calepin](https://github.com/wikistat/Ateliers-Big-Data/blob/master/3-MovieLens/Atelier-MovieLens-softImpute.ipynb). La version de [NMF](http://wikistat.fr/pdf/st-m-explo-nmf.pdf) de [MLlib de Spark](http://spark.apache.org/docs/latest/api/python/pyspark.mllib.html#pyspark.mllib.recommendation.ALS) autorise permet également la complétion.\n\nEn revanche,la version de NMF incluse dans la librairie [Scikit-learn](http://scikit-learn.org/stable/modules/generated/sklearn.decomposition.NMF.html) traite également des [matrices creuses](http://docs.scipy.org/doc/scipy/reference/sparse.html) mais le critère (moindres carrés) optimisé considère les \"0\" comme des notes nulles, pas comme des données manquantes. *Elle n'est pas adaptée au problème de complétion*, contrairement à celle de MLliB. Il faudrait sans doute utiliser la librairie [nonnegfac](https://github.com/kimjingu/nonnegfac-python) en Python de [Kim et al. (2014)](http://link.springer.com/content/pdf/10.1007%2Fs10898-013-0035-4.pdf); **à tester**!\n\nDans la première partie, le plus petit fichier est partagé en trois échantillons: apprentissage, validation et test; l'optimisation du rang de la factorisation (nombre de facteurs latents) est réalisée par minimisation de l'erreur estimée sur l'échantillon de validation.\n\nEnsuite le plus gros fichier est utilisé pour évaluer l'impact de la taille de la base d'apprentissage.", "_____no_output_____" ], [ "## 2 Importation des données en HDFS\nLes données doivent être stockées à un emplacement accessibles de tous les noeuds du cluster pour permettre la construction de la base de données réparties (RDD). Dans une utilisation monoposte (*standalone*) de *Spark*, elles sont simplement chargées dans le répertoire courant. ", "_____no_output_____" ] ], [ [ "sc", "_____no_output_____" ], [ "# Chargement des fichiers si ce n'est déjà fait\n#Renseignez ici le dossier où vous souhaitez stocker le fichier téléchargé.\nDATA_PATH=\"\" \nimport urllib.request\n# fichier réduit\nf = urllib.request.urlretrieve(\"http://www.math.univ-toulouse.fr/~besse/Wikistat/data/ml-ratings100k.csv\",DATA_PATH+\"ml-ratings100k.csv\")", "_____no_output_____" ] ], [ [ "Les données sont lues comme une seule ligne de texte avant d'être restructurées au bon format d'une *matrice creuse* à savoir une liste de triplets contenant les indices de ligne, de colonne et la note pour les seules valeurs renseignées.", "_____no_output_____" ] ], [ [ "# Importer les données au format texte dans un RDD\n\nsmall_ratings_raw_data = sc.textFile(DATA_PATH+\"ml-ratings100k.csv\")\n\n# Identifier et afficher la première ligne\nsmall_ratings_raw_data_header = small_ratings_raw_data.take(1)[0]\nprint(small_ratings_raw_data_header)\n\n# Create RDD without header\nall_lines = small_ratings_raw_data.filter(lambda l : l!=small_ratings_raw_data_header)", "_____no_output_____" ], [ "# Séparer les champs (user, item, note) dans un nouveau RDD\nfrom pyspark.sql import Row\nsplit_lines = all_lines.map(lambda l : l.split(\",\"))\nratingsRDD = split_lines.map(lambda p: Row(user=int(p[0]), item=int(p[1]),\n rating=float(p[2]), timestamp=int(p[3])))\n\n# .cache() : le RDD est conservé en mémoire une fois traité\nratingsRDD.cache()\n\n# Display the two first rows\nratingsRDD.take(2)", "_____no_output_____" ], [ "# Convert RDD to DataFrame\nratingsDF = spark.createDataFrame(ratingsRDD)\nratingsDF.take(2)", "_____no_output_____" ] ], [ [ "## 3. Optimisation du rang sur l'échantillon 10k\nLe fichier comporte 10 000 évaluations croisant les avis de mille utilisateurs sur les films qu'ils ont vus parmi 1700.", "_____no_output_____" ], [ "### 3.1 Constitution des échantillons", "_____no_output_____" ], [ "Séparation aléatoire en trois échantillons apprentissage, validation et test. Le paramètre de rang est optimisé en minimisant l'estimaiton de l'erreur sur l'échantillon test. Cette stratégie, plutôt qu'ue validation croisée est plus adaptée à des données massives.\n", "_____no_output_____" ] ], [ [ "tauxTrain=0.6\ntauxVal=0.2\ntauxTes=0.2\n# Si le total est inférieur à 1, les données sont sous-échantillonnées.\n(trainDF, validDF, testDF) = ratingsDF.randomSplit([tauxTrain, tauxVal, tauxTes])\n# validation et test à prédire, sans les notes\nvalidDF_P = validDF.select(\"user\", \"item\")\ntestDF_P = testDF.select(\"user\", \"item\")", "_____no_output_____" ], [ "trainDF.take(2), validDF_P.take(2), testDF_P.take(2)", "_____no_output_____" ] ], [ [ "### 3.2 Optimisation du rang de la NMF", "_____no_output_____" ], [ "L'erreur d'imputation des données, donc de recommandation, est estimée sur l'échantillon de validation pour différentes valeurs (grille) du rang de la factorisation matricielle. \n\nIl faudrait en principe aussi optimiser la valeur du paramètre de pénalisation pris à 0.1 par défaut.\n\n*Point important:* l'erreur d'ajustement de la factorisation ne prend en compte que les valeurs listées dans la matrice creuses, pas les \"0\" qui sont des données manquantes.", "_____no_output_____" ] ], [ [ "from pyspark.ml.recommendation import ALS\nimport math\nimport collections\n# Initialisation du générateur\nseed = 5\n# Nombre max d'itérations (ALS)\nmaxIter = 10\n# Régularisation L1; à optimiser également\nregularization_parameter = 0.1\n# Choix d'une grille pour les valeurs du rang à optimiser\nranks = [4, 8, 12]\n\n#Initialisation variable \n# création d'un dictionaire pour stocker l'erreur par rang testé\nerrors = collections.defaultdict(float)\ntolerance = 0.02\nmin_error = float('inf')\nbest_rank = -1\nbest_iteration = -1", "_____no_output_____" ], [ "from pyspark.ml.evaluation import RegressionEvaluator\n\nfor rank in ranks:\n als = ALS( rank=rank, seed=seed, maxIter=maxIter,\n regParam=regularization_parameter)\n model = als.fit(trainDF)\n # Prévision de l'échantillon de validation\n predDF = model.transform(validDF).select(\"prediction\",\"rating\")\n #Remove unpredicter row due to no-presence of user in the train dataset\n pred_without_naDF = predDF.na.drop()\n # Calcul du RMSE\n evaluator = RegressionEvaluator(metricName=\"rmse\", labelCol=\"rating\",\n predictionCol=\"prediction\")\n rmse = evaluator.evaluate(pred_without_naDF)\n print(\"Root-mean-square error for rank %d = \"%rank + str(rmse))\n errors[rank] = rmse\n if rmse < min_error:\n min_error = rmse\n best_rank = rank\n# Meilleure solution\nprint('Rang optimal: %s' % best_rank)", "_____no_output_____" ] ], [ [ "### 3.3 Résultats et test", "_____no_output_____" ] ], [ [ "# Quelques prévisions\npred_without_naDF.take(3)", "_____no_output_____" ] ], [ [ "Prévision finale de l'échantillon test.", "_____no_output_____" ] ], [ [ "#On concatane la DataFrame Train et Validatin\ntrainValidDF = trainDF.union(validDF)\n# On crée un model avec le nouveau Dataframe complété d'apprentissage et le rank fixé à la valeur optimal \nals = ALS( rank=best_rank, seed=seed, maxIter=maxIter,\n regParam=regularization_parameter)\nmodel = als.fit(trainValidDF)\n#Prediction sur la DataFrame Test\ntestDF = model.transform(testDF).select(\"prediction\",\"rating\")\n#Remove unpredicter row due to no-presence of user in the trai dataset\npred_without_naDF = predDF.na.drop()\n# Calcul du RMSE\nevaluator = RegressionEvaluator(metricName=\"rmse\", labelCol=\"rating\",\n predictionCol=\"prediction\")\nrmse = evaluator.evaluate(pred_without_naDF)\nprint(\"Root-mean-square error for rank %d = \"%best_rank + str(rmse))\n\n", "_____no_output_____" ] ], [ [ "## 3 Analyse du fichier complet", "_____no_output_____" ], [ "MovieLens propose un plus gros fichier avec 20M de notes (138000 utilisateurs, 27000 films). Ce fichier est utilisé pour extraire un fichier test de deux millions de notes à reconstruire. Les paramètres précédemment optimisés, ils pourraient sans doute l'être mieux, sont appliqués pour une succesion d'estimation / prévision avec une taille croissante de l'échantillon d'apprentissage. Il aurait été plus élégant d'automatiser le travail dans une boucle mais lorsque les données sont les plus volumineuses des comportement mal contrôlés de Spark peuvent provoquer des plantages par défaut de mémoire.", "_____no_output_____" ], [ "### 3.1 Lecture des données", "_____no_output_____" ], [ "Le fichier est prétraité de manière analogue.", "_____no_output_____" ] ], [ [ "# Chargement des fichiers si ce n'est déjà fait\nimport urllib.request\n# fichier complet mais compressé\nf = urllib.request.urlretrieve(\"http://www.math.univ-toulouse.fr/~besse/Wikistat/data/ml-ratings20M.zip\",DATA_PATH+\"ml-ratings20M.zip\")", "_____no_output_____" ], [ "#Unzip downloaded file\nimport zipfile\nzip_ref = zipfile.ZipFile(DATA_PATH+\"ml-ratings20M.zip\", 'r')\nzip_ref.extractall(DATA_PATH)\nzip_ref.close()", "_____no_output_____" ], [ "# Importer les données au format texte dans un RDD\nratings_raw_data = sc.textFile(DATA_PATH+\"ratings20M.csv\")\n# Identifier et afficher la première ligne\nratings_raw_data_header = ratings_raw_data.take(1)[0]\nratings_raw_data_header\n\n# Create RDD without header\nall_lines = ratings_raw_data.filter(lambda l : l!=ratings_raw_data_header)\n", "_____no_output_____" ], [ "# Séparer les champs (user, item, note) dans un nouveau RDD\nsplit_lines = all_lines.map(lambda l : l.split(\",\"))\nratingsRDD = split_lines.map(lambda p: Row(user=int(p[0]), item=int(p[1]),\n rating=float(p[2]), timestamp=int(p[3])))\n\n# Display the two first rows\nratingsRDD.take(2)", "_____no_output_____" ], [ "# Convert RDD to DataFrame\nratingsDF = spark.createDataFrame(ratingsRDD)\nratingsDF.take(2)", "_____no_output_____" ] ], [ [ "### 3.2 Echantillonnage", "_____no_output_____" ], [ "Extraction de l'échantillon test et éventuellement sous-échantillonnage de l'échantillon d'apprentissage. ", "_____no_output_____" ] ], [ [ "tauxTest=0.1\n# Si le total est inférieur à 1, les données sont sous-échantillonnées.\n(trainTotDF, testDF) = ratingsDF.randomSplit([1-tauxTest, tauxTest])", "_____no_output_____" ], [ "# Sous-échantillonnage de l'apprentissage permettant de \n# tester pour des tailles croissantes de cet échantillon\ntauxEch=0.2\n(trainDF, DropData) = trainTotDF.randomSplit([tauxEch, 1-tauxEch])", "_____no_output_____" ], [ "testDF.take(2), trainDF.take(2)", "_____no_output_____" ] ], [ [ "### 3.3 Estimation du modèle", "_____no_output_____" ], [ "Le modèle est estimé en utilisant les valeurs des paramètres obtenues dans l'étape précédente.", "_____no_output_____" ] ], [ [ "import time\ntime_start=time.time()\n# Initialisation du générateur\nseed = 5\n# Nombre max d'itérations (ALS)\nmaxIter = 10\n# Régularisation L1 (valeur par défaut)\nregularization_parameter = 0.1\nbest_rank = 8\n# Estimation pour chaque valeur de rang\nals = ALS(rank=rank, seed=seed, maxIter=maxIter,\n regParam=regularization_parameter)\nmodel = als.fit(trainDF)\ntime_end=time.time()\ntime_als=(time_end - time_start)\nprint(\"ALS prend %d s\" %(time_als)) ", "_____no_output_____" ] ], [ [ "### 3.4 Prévision de l'échantillon test et erreur", "_____no_output_____" ] ], [ [ "# Prévision de l'échantillon de validation\npredDF = model.transform(testDF).select(\"prediction\",\"rating\")\n#Remove unpredicter row due to no-presence of user in the train dataset\npred_without_naDF = predDF.na.drop()\n# Calcul du RMSE\nevaluator = RegressionEvaluator(metricName=\"rmse\", labelCol=\"rating\",\n predictionCol=\"prediction\")\nrmse = evaluator.evaluate(pred_without_naDF)\nprint(\"Root-mean-square error for rank %d = \"%best_rank + str(rmse))", "_____no_output_____" ], [ "trainDF.count()", "_____no_output_____" ] ], [ [ "Quelques résultats montrant l'évolution du temps de calcul et de l'erreur de prévision en fonction de la taille de l'échantillon d'apprentissage. Attention, il est probable que la valeur des paramètres optimaux dépendent de la taille de l'échantillon d'apprentissage.\n\nTaille | Temps(s) | RMSE\n-------|-------|------\n217439 | 70 | 1.65\n1029416| 73 | 1.06\n2059855| 72 | 1.05\n4119486| 89 | 0.88\n6176085| 99 | 0.85\n10301909| 117 | 0.83\n12361034| 125 | 0.83\n14414907| 137 | 0.82\n16474087| 148 | 0.818\n18538142| 190 | 0.816\n20596263| 166 | 0.82\n", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# Experiments comparing the performance of traditional pooling operations and entropy pooling within a shallow neural network and Lenet. The experiments use cifar10 and cifar100.", "_____no_output_____" ] ], [ [ "%matplotlib inline", "_____no_output_____" ], [ "import torch\nimport torchvision\nimport torchvision.transforms as transforms", "_____no_output_____" ], [ "transform = transforms.Compose(\n [transforms.ToTensor(),\n transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])\n\ntrainset = torchvision.datasets.CIFAR100(root='./data', train=True,\n download=True, transform=transform)\ntrainloader = torch.utils.data.DataLoader(trainset, batch_size=4,\n shuffle=True, num_workers=8)\n\ntestset = torchvision.datasets.CIFAR100(root='./data', train=False,\n download=True, transform=transform)\ntestloader = torch.utils.data.DataLoader(testset, batch_size=4,\n shuffle=False, num_workers=8)\n\nclasses = ('plane', 'car', 'bird', 'cat',\n 'deer', 'dog', 'frog', 'horse', 'ship', 'truck')", "_____no_output_____" ], [ "import math\nimport torch\nimport torch.nn as nn\nimport torch.nn.functional as F\nfrom torch.nn.modules.utils import _pair, _quadruple\nimport time\nfrom skimage.measure import shannon_entropy\nfrom scipy import stats\n\nfrom torch.nn.modules.utils import _pair, _quadruple\nimport time\nfrom skimage.measure import shannon_entropy\nfrom scipy import stats\nimport numpy as np\n\nclass EntropyPool2d(nn.Module):\n\n def __init__(self, kernel_size=3, stride=1, padding=0, same=False, entr='high'):\n super(EntropyPool2d, self).__init__()\n self.k = _pair(kernel_size)\n self.stride = _pair(stride)\n self.padding = _quadruple(padding) # convert to l, r, t, b\n self.same = same\n self.entr = entr\n\n def _padding(self, x):\n if self.same:\n ih, iw = x.size()[2:]\n if ih % self.stride[0] == 0:\n ph = max(self.k[0] - self.stride[0], 0)\n else:\n ph = max(self.k[0] - (ih % self.stride[0]), 0)\n if iw % self.stride[1] == 0:\n pw = max(self.k[1] - self.stride[1], 0)\n else:\n pw = max(self.k[1] - (iw % self.stride[1]), 0)\n pl = pw // 2\n pr = pw - pl\n pt = ph // 2\n pb = ph - pt\n padding = (pl, pr, pt, pb)\n else:\n padding = self.padding\n return padding\n \n def forward(self, x):\n # using existing pytorch functions and tensor ops so that we get autograd, \n # would likely be more efficient to implement from scratch at C/Cuda level\n start = time.time()\n x = F.pad(x, self._padding(x), mode='reflect')\n x_detached = x.cpu().detach()\n x_unique, x_indices, x_inverse, x_counts = np.unique(x_detached,\n return_index=True, \n return_inverse=True, \n return_counts=True) \n freq = torch.FloatTensor([x_counts[i] / len(x_inverse) for i in x_inverse]).cuda()\n x_probs = freq.view(x.shape) \n x_probs = x_probs.unfold(2, self.k[0], self.stride[0]).unfold(3, self.k[1], self.stride[1])\n x_probs = x_probs.contiguous().view(x_probs.size()[:4] + (-1,))\n if self.entr is 'high':\n x_probs, indices = torch.min(x_probs.cuda(), dim=-1)\n elif self.entr is 'low':\n x_probs, indices = torch.max(x_probs.cuda(), dim=-1)\n else:\n raise Exception('Unknown entropy mode: {}'.format(self.entr))\n \n x = x.unfold(2, self.k[0], self.stride[0]).unfold(3, self.k[1], self.stride[1])\n x = x.contiguous().view(x.size()[:4] + (-1,))\n indices = indices.view(indices.size() + (-1,))\n pool = torch.gather(input=x, dim=-1, index=indices)\n \n return pool.squeeze(-1)\n ", "_____no_output_____" ], [ "import torch.nn as nn\nimport torch.nn.functional as F\nimport torch.optim as optim\nimport time\nfrom sklearn.metrics import f1_score\n\nMAX = 'max'\nAVG = 'avg'\nHIGH_ENTROPY = 'high_entr'\nLOW_ENTROPY = 'low_entr'\n\nclass Net1Pool(nn.Module):\n def __init__(self, num_classes=10, pooling=MAX):\n super(Net1Pool, self).__init__()\n self.conv1 = nn.Conv2d(3, 30, 5)\n \n if pooling is MAX:\n self.pool = nn.MaxPool2d(2, 2)\n elif pooling is AVG:\n self.pool = nn.AvgPool2d(2, 2)\n elif pooling is HIGH_ENTROPY:\n self.pool = EntropyPool2d(2, 2, entr='high')\n elif pooling is LOW_ENTROPY:\n self.pool = EntropyPool2d(2, 2, entr='low')\n \n self.fc0 = nn.Linear(30 * 14 * 14, num_classes)\n\n def forward(self, x):\n x = self.pool(F.relu(self.conv1(x)))\n x = x.view(-1, 30 * 14 * 14)\n x = F.relu(self.fc0(x))\n return x\n\n\nclass Net2Pool(nn.Module):\n def __init__(self, num_classes=10, pooling=MAX):\n super(Net2Pool, self).__init__()\n self.conv1 = nn.Conv2d(3, 50, 5, 1)\n self.conv2 = nn.Conv2d(50, 50, 5, 1)\n \n if pooling is MAX:\n self.pool = nn.MaxPool2d(2, 2)\n elif pooling is AVG:\n self.pool = nn.AvgPool2d(2, 2)\n elif pooling is HIGH_ENTROPY:\n self.pool = EntropyPool2d(2, 2, entr='high')\n elif pooling is LOW_ENTROPY:\n self.pool = EntropyPool2d(2, 2, entr='low')\n \n self.fc1 = nn.Linear(5*5*50, 500)\n self.fc2 = nn.Linear(500, num_classes)\n\n def forward(self, x):\n x = F.relu(self.conv1(x))\n x = self.pool(x)\n x = F.relu(self.conv2(x))\n x = self.pool(x)\n\n x = x.view(-1, 5*5*50)\n x = F.relu(self.fc1(x))\n x = self.fc2(x)\n return x\n\ndef configure_net(net, device):\n net.to(device)\n criterion = nn.CrossEntropyLoss()\n optimizer = optim.SGD(net.parameters(), lr=0.001, momentum=0.9)\n return net, optimizer, criterion\n\ndef train(net, optimizer, criterion, trainloader, device, epochs=10, logging=2000):\n for epoch in range(epochs): \n running_loss = 0.0\n for i, data in enumerate(trainloader, 0):\n start = time.time()\n inputs, labels = data\n inputs, labels = inputs.to(device), labels.to(device)\n \n optimizer.zero_grad()\n outputs = net(inputs)\n loss = criterion(outputs, labels)\n loss.backward()\n optimizer.step()\n\n running_loss += loss.item()\n if i % logging == logging - 1: \n print('[%d, %5d] loss: %.3f duration: %.5f' %\n (epoch + 1, i + 1, running_loss / logging, time.time() - start))\n running_loss = 0.0\n\n print('Finished Training')\n \ndef test(net, testloader, device):\n correct = 0\n total = 0\n predictions = []\n l = []\n with torch.no_grad():\n for data in testloader:\n images, labels = data\n images, labels = images.to(device), labels.to(device)\n\n outputs = net(images)\n _, predicted = torch.max(outputs.data, 1)\n total += labels.size(0)\n correct += (predicted == labels).sum().item()\n predictions.extend(predicted.cpu().numpy())\n l.extend(labels.cpu().numpy())\n\n\n print('Accuracy: {}'.format(100 * correct / total))\n", "_____no_output_____" ], [ "epochs = 10\nlogging = 15000\nnum_classes = 100\n", "_____no_output_____" ], [ "print('- - - - - - - - -- - - - 2 pool - - - - - - - - - - - - - - - -')\nprint('- - - - - - - - -- - - - MAX - - - - - - - - - - - - - - - -')\n\ndevice = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\nnet, optimizer, criterion = configure_net(Net2Pool(num_classes=num_classes, pooling=MAX), device)\ntrain(net, optimizer, criterion, trainloader, device, epochs=epochs, logging=logging)\ntest(net, testloader, device)\n\nprint('- - - - - - - - -- - - - AVG - - - - - - - - - - - - - - - -')\n\nnet, optimizer, criterion = configure_net(Net2Pool(num_classes=num_classes, pooling=AVG), device)\ntrain(net, optimizer, criterion, trainloader, device, epochs=epochs, logging=logging)\ntest(net, testloader, device)\n\nprint('- - - - - - - - -- - - - HIGH - - - - - - - - - - - - - - - -')\n\nnet, optimizer, criterion = configure_net(Net2Pool(num_classes=num_classes, pooling=HIGH_ENTROPY), device)\ntrain(net, optimizer, criterion, trainloader, device, epochs=epochs, logging=logging)\ntest(net, testloader, device)\n\nprint('- - - - - - - - -- - - - LOW - - - - - - - - - - - - - - - -')\n\nnet, optimizer, criterion = configure_net(Net2Pool(num_classes=num_classes, pooling=LOW_ENTROPY), device)\ntrain(net, optimizer, criterion, trainloader, device, epochs=epochs, logging=logging)\ntest(net, testloader, device)", "_____no_output_____" ], [ "print('- - - - - - - - -- - - - 1 pool - - - - - - - - - - - - - - - -')\nprint('- - - - - - - - -- - - - MAX - - - - - - - - - - - - - - - -')\n\ndevice = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\nnet, optimizer, criterion = configure_net(Net1Pool(num_classes=num_classes, pooling=MAX), device)\ntrain(net, optimizer, criterion, trainloader, device, epochs=epochs, logging=logging)\ntest(net, testloader, device)\n\nprint('- - - - - - - - -- - - - AVG - - - - - - - - - - - - - - - -')\n\nnet, optimizer, criterion = configure_net(Net1Pool(num_classes=num_classes, pooling=AVG), device)\ntrain(net, optimizer, criterion, trainloader, device, epochs=epochs, logging=logging)\ntest(net, testloader, device)\n\nprint('- - - - - - - - -- - - - HIGH - - - - - - - - - - - - - - - -')\n\nnet, optimizer, criterion = configure_net(Net1Pool(num_classes=num_classes, pooling=HIGH_ENTROPY), device)\ntrain(net, optimizer, criterion, trainloader, device, epochs=epochs, logging=logging)\ntest(net, testloader, device)\n\nprint('- - - - - - - - -- - - - LOW - - - - - - - - - - - - - - - -')\n\nnet, optimizer, criterion = configure_net(Net1Pool(num_classes=num_classes, pooling=LOW_ENTROPY), device)\ntrain(net, optimizer, criterion, trainloader, device, epochs=epochs, logging=logging)\ntest(net, testloader, device)", "_____no_output_____" ] ] ]

[ "markdown", "code" ]

[ [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "<a href=\"https://colab.research.google.com/github/AIWintermuteAI/aXeleRate/blob/dev/resources/aXeleRate_mark_detector.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>", "_____no_output_____" ], [ "## M.A.R.K. Detection model Training and Inference\n\nIn this notebook we will use axelerate, Keras-based framework for AI on the edge, to quickly setup model training and then after training session is completed convert it to .tflite and .kmodel formats.\n\nFirst, let's take care of some administrative details. \n\n1) Before we do anything, make sure you have choosen GPU as Runtime type (in Runtime - > Change Runtime type).\n\n2) We need to mount Google Drive for saving our model checkpoints and final converted model(s). Press on Mount Google Drive button in Files tab on your left. \n\nIn the next cell we clone axelerate Github repository and import it. \n\n**It is possible to use pip install or python setup.py install, but in that case you will need to restart the enironment.** Since I'm trying to make the process as streamlined as possibile I'm using sys.path.append for import.", "_____no_output_____" ] ], [ [ "%load_ext tensorboard\n#we need imgaug 0.4 for image augmentations to work properly, see https://stackoverflow.com/questions/62580797/in-colab-doing-image-data-augmentation-with-imgaug-is-not-working-as-intended\n!pip uninstall -y imgaug && pip uninstall -y albumentations && pip install imgaug==0.4\n!git clone https://github.com/AIWintermuteAI/aXeleRate.git\nimport sys\nsys.path.append('/content/aXeleRate')\nfrom axelerate import setup_training, setup_inference", "_____no_output_____" ] ], [ [ "At this step you typically need to get the dataset. You can use !wget command to download it from somewhere on the Internet or !cp to copy from My Drive as in this example\n```\n!cp -r /content/drive/'My Drive'/pascal_20_segmentation.zip .\n!unzip --qq pascal_20_segmentation.zip\n```\nDataset preparation and postprocessing are discussed in the article here:\n\nThe annotation tool I use is LabelImg\nhttps://github.com/tzutalin/labelImg\n\nLet's visualize our detection model test dataset. There are images in validation folder with corresponding annotations in PASCAL-VOC format in validation annotations folder.\n", "_____no_output_____" ] ], [ [ "%matplotlib inline\n!gdown https://drive.google.com/uc?id=1s2h6DI_1tHpLoUWRc_SavvMF9jYG8XSi #dataset\n!gdown https://drive.google.com/uc?id=1-bDRZ9Z2T81SfwhHEfZIMFG7FtMQ5ZiZ #pre-trained model\n\n!unzip --qq mark_dataset.zip\n\nfrom axelerate.networks.common_utils.augment import visualize_detection_dataset\n\nvisualize_detection_dataset(img_folder='mark_detection/imgs_validation', ann_folder='mark_detection/ann_validation', num_imgs=10, img_size=224, augment=True)", "_____no_output_____" ] ], [ [ "Next step is defining a config dictionary. Most lines are self-explanatory.\n\nType is model frontend - Classifier, Detector or Segnet\n\nArchitecture is model backend (feature extractor) \n\n- Full Yolo\n- Tiny Yolo\n- MobileNet1_0\n- MobileNet7_5 \n- MobileNet5_0 \n- MobileNet2_5 \n- SqueezeNet\n- NASNetMobile\n- DenseNet121\n- ResNet50\n\nFor more information on anchors, please read here\nhttps://github.com/pjreddie/darknet/issues/568\n\nLabels are labels present in your dataset.\nIMPORTANT: Please, list all the labels present in the dataset.\n\nobject_scale determines how much to penalize wrong prediction of confidence of object predictors\n\nno_object_scale determines how much to penalize wrong prediction of confidence of non-object predictors\n\ncoord_scale determines how much to penalize wrong position and size predictions (x, y, w, h)\n\nclass_scale determines how much to penalize wrong class prediction\n\nFor converter type you can choose the following:\n\n'k210', 'tflite_fullint', 'tflite_dynamic', 'edgetpu', 'openvino', 'onnx'\n", "_____no_output_____" ], [ "## Parameters for Person Detection\n\nK210, which is where we will run the network, has constrained memory (5.5 RAM) available, so with Micropython firmware, the largest model you can run is about 2 MB, which limits our architecture choice to Tiny Yolo, MobileNet(up to 0.75 alpha) and SqueezeNet. Out of these 3 architectures, only one comes with pre-trained model - MobileNet. So, to save the training time we will use Mobilenet with alpha 0.75, which has ... parameters. For objects that do not have that much variety, you can use MobileNet with lower alpha, down to 0.25.", "_____no_output_____" ] ], [ [ "config = {\n \"model\":{\n \"type\": \"Detector\",\n \"architecture\": \"MobileNet5_0\",\n \"input_size\": 224,\n \"anchors\": [0.57273, 0.677385, 1.87446, 2.06253, 3.33843, 5.47434, 7.88282, 3.52778, 9.77052, 9.16828],\n \"labels\": [\"mark\"],\n \"coord_scale\" : \t\t1.0,\n \"class_scale\" : \t\t1.0,\n \"object_scale\" : \t\t5.0,\n \"no_object_scale\" : \t1.0\n },\n \"weights\" : {\n \"full\": \t\t\t\t\"\",\n \"backend\": \t\t \"imagenet\"\n },\n \"train\" : {\n \"actual_epoch\": 50,\n \"train_image_folder\": \"mark_detection/imgs\",\n \"train_annot_folder\": \"mark_detection/ann\",\n \"train_times\": 1,\n \"valid_image_folder\": \"mark_detection/imgs_validation\",\n \"valid_annot_folder\": \"mark_detection/ann_validation\",\n \"valid_times\": 1,\n \"valid_metric\": \"mAP\",\n \"batch_size\": 32,\n \"learning_rate\": 1e-3,\n \"saved_folder\": \t\tF\"/content/drive/MyDrive/mark_detector\",\n \"first_trainable_layer\": \"\",\n \"augumentation\":\t\t\t\tTrue,\n \"is_only_detect\" : \t\tFalse\n },\n \"converter\" : {\n \"type\": \t\t\t\t[\"k210\",\"tflite\"]\n }\n }", "_____no_output_____" ] ], [ [ "Let's check what GPU we have been assigned in this Colab session, if any.", "_____no_output_____" ] ], [ [ "from tensorflow.python.client import device_lib\ndevice_lib.list_local_devices()", "_____no_output_____" ] ], [ [ "Also, let's open Tensorboard, where we will be able to watch model training progress in real time. Training and validation logs also will be saved in project folder.\nSince there are no logs before we start the training, tensorboard will be empty. Refresh it after first epoch.", "_____no_output_____" ] ], [ [ "%tensorboard --logdir logs", "_____no_output_____" ] ], [ [ "Finally we start the training by passing config dictionary we have defined earlier to setup_training function. The function will start the training with Checkpoint, Reduce Learning Rate on Plateau and Early Stopping callbacks. After the training has stopped, it will convert the best model into the format you have specified in config and save it to the project folder.", "_____no_output_____" ] ], [ [ "from keras import backend as K \nK.clear_session()\nmodel_path = setup_training(config_dict=config)", "_____no_output_____" ] ], [ [ "After training it is good to check the actual perfomance of your model by doing inference on your validation dataset and visualizing results. This is exactly what next block does. Obviously since our model has only trained on a few images the results are far from stellar, but if you have a good dataset, you'll have better results.", "_____no_output_____" ] ], [ [ "from keras import backend as K \nK.clear_session()\nsetup_inference(config, model_path)", "_____no_output_____" ] ], [ [ "My end results are:\n\n{'fscore': 0.942528735632184, 'precision': 0.9318181818181818, 'recall': 0.9534883720930233}\n\n**You can obtain these results by loading a pre-trained model.**\n\nGood luck and happy training! Have a look at these articles, that would allow you to get the most of Google Colab or connect to local runtime if there are no GPUs available;\n\nhttps://medium.com/@oribarel/getting-the-most-out-of-your-google-colab-2b0585f82403\n\nhttps://research.google.com/colaboratory/local-runtimes.html", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Example of how to use the CGMFtk package", "_____no_output_____" ] ], [ [ "# Table of Contents\n1. [Import Modules](#import)\n2. [Read the History File](#read)\n3. [Summary Table](#table)\n4. [Histogram Fission Fragment Properties](#ffHistograms)\n5. [Correlated Observables](#correlations)\n6. [Neutron Properties](#neutrons)\n7. [Gamma Properties](#gammas)\n8. [Gamma-ray Timing Information](#timing)\n9. [Angular Correlations](#angles)", "_____no_output_____" ], [ "### 1. import modules for the notebook\n<a id='import'></a>", "_____no_output_____" ] ], [ [ "import numpy as np\nimport os\nimport matplotlib.pyplot as plt\nfrom CGMFtk import histories as fh", "_____no_output_____" ], [ "# also define some plotting features\nimport matplotlib as mpl\nmpl.rcParams['font.size'] = 12\nmpl.rcParams['font.family'] = 'Helvetica','serif'\nmpl.rcParams['font.weight'] = 'normal'\nmpl.rcParams['axes.labelsize'] = 18.\nmpl.rcParams['xtick.labelsize'] = 18.\nmpl.rcParams['ytick.labelsize'] = 18.\nmpl.rcParams['lines.linewidth'] = 2.\nmpl.rcParams['xtick.major.pad'] = '10'\nmpl.rcParams['ytick.major.pad'] = '10'\nmpl.rcParams['image.cmap'] = 'BuPu'", "_____no_output_____" ], [ "# define a working directory where the history files are stored\nworkdir = './'\nhistFile = 'histories.out'\ntimeFile = 'histories.out'\nyeildFile = 'yeilds.cgmf.0'\n\nnevents = int(1E6)", "_____no_output_____" ] ], [ [ "### 2. Read CGMF history file\n<a id='read'></a>", "_____no_output_____" ] ], [ [ "# run Cf-252 sf with a single OM param file (#42)\ndirectory = \"/home/beykyle/db/projects/OM/KDOMPuq/KDUQSamples\"\nfor filename in os.scandir(directory):\n if filename.is_file() and \"42\" in filename.path:\n #if filename.is_file():\n print(filename.path)\n os.system(\"mpirun -np 8 --use-hwthread-cpus cgmf.mpi.x -t -1 -i 98252 -e 0.0 -n 100 -o\" + filename.path)\n os.system(\"cat histories.cgmf.* > histories.out\")\n os.system(\"rm histories.cgmf.*\")\n \n ", "/home/beykyle/db/projects/OM/KDOMPuq/KDUQSamples/42.json\n" ], [ "print(\"Analyzing histories\")\nhist = fh.Histories(workdir + histFile, nevents=nevents)\n# print the number of events in the file\nprint ('This file contains ',str(hist.getNumberEvents()),' events and ',str(hist.getNumberFragments()),' fission fragments')", "Analyzing histories\nWARNING\nYou asked for 1000000 events and there are only 800 in this history file\nThis file contains 800 events and 1600 fission fragments\n" ] ], [ [ "With the option 'nevents', the number of fission events that are read can be specified:", "_____no_output_____" ], [ "hist = fh.Histories('92235_1MeV.cgmf',nevents=5000)", "_____no_output_____" ], [ "### 3. Summary Table\n<a id='table'></a>", "_____no_output_____" ] ], [ [ "# provide a summary table of the fission events\nhist.summaryTable()", "_____no_output_____" ] ], [ [ "### 4. Fission Fragment Properties\n<a id='ffHistograms'></a>", "_____no_output_____" ], [ "With the histogram function from matplotlib, we can easily plot distributions of the fission fragment characteristics", "_____no_output_____" ] ], [ [ "# plot the distributions of the fission fragments\nA = hist.getA() # get A of all fragments\nAL = hist.getALF() # get A of light fragments\nAH = hist.getAHF() # get A of heavy fragments\nfig = plt.figure(figsize=(8,6))\nbins = np.arange(min(A),max(A))\nh,b = np.histogram(A,bins=bins,density=True)\nplt.plot(b[:-1],h,'-o')\nplt.xlabel('Mass (A)')\nplt.ylabel('Frequency')\nplt.show()\n\nZ = hist.getZ()\nZL = hist.getZLF()\nZH = hist.getZHF()\nfig = plt.figure(figsize=(8,6))\nbins = np.arange(min(Z),max(Z))\nh,b = np.histogram(Z,bins=bins,density=True)\nplt.plot(b[:-1],h,'-o')\nplt.xlabel('Charge (Z)')\nplt.ylabel('Frequency')\nplt.show()\n\nfig = plt.figure(figsize=(8,6))\nTKEpre = hist.getTKEpre() # TKE before neutron emission\nTKEpost = hist.getTKEpost() # TKE after neutron emission\nbins = np.arange(min(TKEpre),max(TKEpre))\nh,b = np.histogram(TKEpre,bins=bins,density=True)\nplt.plot(0.5*(b[:-1]+b[1:]),h,'-o',label='Before neutron emission')\nbins = np.arange(min(TKEpost),max(TKEpost))\nh,b = np.histogram(TKEpost,bins=bins,density=True)\nplt.plot(0.5*(b[:-1]+b[1:]),h,'-o',label='After neutron emission')\nplt.legend()\nplt.xlabel('Total Kinetic Energy (MeV)')\nplt.ylabel('Frequency')\nplt.show()", "_____no_output_____" ] ], [ [ "With the 2D histogram feature, we can see correlations between the calculated features from CGMF", "_____no_output_____" ] ], [ [ "TKEpre = hist.getTKEpre()\nTXE = hist.getTXE()\nbx = np.arange(min(TKEpre),max(TKEpre))\nby = np.arange(min(TXE),max(TXE))\nfig = plt.figure(figsize=(8,6))\nplt.hist2d(TKEpre,TXE,bins=(bx,by),density=True)\nplt.xlabel('Total Kinetic Energy (MeV)')\nplt.ylabel('Total Excitation Energy (MeV)')\nplt.colorbar()\nplt.show()", "_____no_output_____" ] ], [ [ "### 5. Correlated Observables\n<a id='correlations'></a>", "_____no_output_____" ], [ "Many observables within fission are correlated with one another. Sometimes, these are best visualized as two-dimensional histograms as in the TKE-TXE plot directly above. Other times, it is helpful to plot certain observables as a function of mass or TKE. There are routines within CGMFtk that easily construct those, as demonstrated here:", "_____no_output_____" ] ], [ [ "# nubar as a function of mass\n## nubarg, excitation energy, kinetic energy (pre), and spin are available as a function of mass\nnubarA = hist.nubarA()\nTKEA = hist.TKEA()\nfig = plt.figure(figsize=(16,6))\nplt.subplot(121)\nplt.plot(nubarA[0],nubarA[1],'ko')\nplt.xlabel('Mass (u)')\nplt.ylabel(r'$\\overline{\\nu}$')\nplt.subplot(122)\nplt.plot(TKEA[0],TKEA[1],'ko')\nplt.xlabel('Mass (u)')\nplt.ylabel(r'Total Kinetic Energy (MeV)')\nplt.tight_layout()\nplt.show()", "_____no_output_____" ] ], [ [ "### 6. Neutron properties\n<a id='neutrons'></a>", "_____no_output_____" ] ], [ [ "# construct and plot the neutron multiplicity distribution\nnu,pnu = hist.Pnu()\nfig = plt.figure(figsize=(8,6))\nplt.plot(nu,pnu,'k*--',markersize=10)\nplt.xlabel(r'$\\nu$')\nplt.ylabel(r'P($\\nu$)')\nplt.show()", "_____no_output_____" ], [ "# construct and plot the prompt neutron spectrum\nfig = plt.figure(figsize=(16,6))\nplt.subplot(121)\nebins,pfns = hist.pfns()\nplt.step(ebins,pfns,where='mid')\nplt.xlim(0,20)\nplt.xlabel('Outgoing neutron energy (MeV)')\nplt.ylabel('PFNS')\nplt.subplot(122)\nplt.step(ebins,pfns,where='mid')\nplt.xlim(0.01,20)\nplt.xscale('log')\nplt.yscale('log')\nplt.xlabel('Outgoing neutron energy (MeV)')\nplt.ylabel('PFNS')\nplt.tight_layout()\nplt.show()", "_____no_output_____" ], [ "# average number of prompt neutrons\nprint ('nubar (per fission event) = ',hist.nubartot())\nprint ('average number of neutrons per fragment = ',hist.nubar())", "nubar (per fission event) = 3.76375\naverage number of neutrons per fragment = 1.881875\n" ], [ "# average neutron energies\nprint ('Neutron energies in the lab:')\nprint ('Average energy of all neutrons = ',hist.meanNeutronElab())\nprint ('Average energy of neutrons from fragments = ',hist.meanNeutronElabFragments())\nprint ('Average energy of neutrons from light fragment = ',hist.meanNeutronElabLF())\nprint ('Average energy of neutrons from heavy fragment = ',hist.meanNeutronElabHF())\nprint (' ')\nprint ('Neutron energies in the center of mass:')\nprint ('Average energy of neutrons from fragments = ',hist.meanNeutronEcmFragments())\nprint ('Average energy of neutrons from light fragment = ',hist.meanNeutronEcmLF())\nprint ('Average energy of neutrons from heavy fragment = ',hist.meanNeutronEcmHF())", "Neutron energies in the lab:\nAverage energy of all neutrons = 2.0337924277648622\nAverage energy of neutrons from fragments = 2.0337924277648622\nAverage energy of neutrons from light fragment = 2.2622832643406268\nAverage energy of neutrons from heavy fragment = 1.7410818181818182\n \nNeutron energies in the center of mass:\nAverage energy of neutrons from fragments = 1.2891006310195947\nAverage energy of neutrons from light fragment = 1.3413187463039622\nAverage energy of neutrons from heavy fragment = 1.2222060606060605\n" ] ], [ [ "Note that the energies are not recorded for the pre-fission neutrons in the center of mass frame", "_____no_output_____" ], [ "### 7. Gamma properties\n<a id='gammas'></a>", "_____no_output_____" ] ], [ [ "# construct and plot the gamma multiplicity distribution\nnug,pnug = hist.Pnug()\nfig = plt.figure(figsize=(8,6))\nplt.plot(nug,pnug,'k*--',markersize=10)\nplt.xlabel(r'$N_\\gamma$')\nplt.ylabel(r'P($N_\\gamma$)')\nplt.show()", "_____no_output_____" ], [ "# construct and plot the prompt neutron spectrum\nfig = plt.figure(figsize=(16,6))\nplt.subplot(121)\nebins,pfgs = hist.pfgs()\nplt.step(ebins,pfgs,where='mid')\nplt.xlim(0,5)\nplt.xlabel(r'Outgoing $\\gamma$ energy (MeV)')\nplt.ylabel('PFGS')\nplt.subplot(122)\nplt.step(ebins,pfgs,where='mid')\nplt.xlim(0.1,5)\nplt.xscale('log')\nplt.yscale('log')\nplt.xlabel(r'Outgoing $\\gamma$ energy (MeV)')\nplt.ylabel('PFGS')\nplt.ylim(1e-2,30)\nplt.tight_layout()\nplt.show()", "_____no_output_____" ], [ "# average number of prompt neutrons\nprint ('nugbar (per fission event) = ',hist.nubargtot())\nprint ('average number of gammas per fragment = ',hist.nubarg())", "nugbar (per fission event) = 9.76625\naverage number of gammas per fragment = 4.883125\n" ], [ "# perform gamma-ray spectroscopy\ngE = 0.2125 # gamma ray at 212.5 keV\ndE = 0.01 # 1% energy resolution\ngspec1 = hist.gammaSpec(gE,dE*gE,post=True)\n\n# calculate the percentage of events for each A/Z\nAs1 = np.unique(gspec1[:,1])\ntotEvents = len(gspec1)\nfracA1 = []\nfor A in As1:\n mask = gspec1[:,1]==A\n fracA1.append(len(gspec1[mask])/totEvents)\nZs1 = np.unique(gspec1[:,0])\nfracZ1 = []\nfor Z in Zs1:\n mask = gspec1[:,0]==Z\n fracZ1.append(len(gspec1[mask])/totEvents)\n\nfig = plt.figure(figsize=(8,6))\nplt.plot(As1,fracA1,'--')\nplt.xlabel('Fission Fragment Mass (A)')\nplt.ylabel('Fraction of Events')\nplt.text(135,0.170,r'$\\epsilon_\\gamma$=212.5 keV',fontsize=18)\nplt.show()\n\nfig = plt.figure(figsize=(8,6))\nplt.plot(Zs1,fracZ1,'--')\nplt.xlabel('Fission Fragment Charge (Z)')\nplt.ylabel('Fraction of Events')\nplt.show()", "_____no_output_____" ], [ "# average neutron energies\nprint ('Gamma energies in the lab:')\nprint ('Average energy of all gammas = ',hist.meanGammaElab())\nprint ('Average energy of gammas from light fragment = ',hist.meanGammaElabLF())\nprint ('Average energy of gammas from heavy fragment = ',hist.meanGammaElabHF())", "Gamma energies in the lab:\nAverage energy of all gammas = 0.7157650070395495\nAverage energy of gammas from light fragment = 0.7317019104946348\nAverage energy of gammas from heavy fragment = 0.7005107715430862\n" ] ], [ [ "Note that in the current version of CGMF, only the doppler shifted lab energies are recorded in the history file", "_____no_output_____" ], [ "### 8. Gamma-ray Timing Information\n<a id='timing'></a>", "_____no_output_____" ], [ "We can also calculate quantities that are related to the time at which the 'late' prompt gamma rays are emitted. When the option -t -1 is included in the run time options for CGMF, these gamma-ray times are printed out in the CGMF history file. The Histories class can read these times based on the header of the CGMF history file.", "_____no_output_____" ] ], [ [ "histTime = fh.Histories(workdir + timeFile, nevents=nevents*2)", "WARNING\nYou asked for 2000000 events and there are only 800 in this history file\n" ], [ "# gamma-ray times can be retrieved through \ngammaAges = histTime.getGammaAges()", "_____no_output_____" ] ], [ [ "The nubargtot function can also be used to construct the average gamma-ray multiplicity per fission event as a function of time. In the call to nubarg() or nubargtot(), timeWindow=True should be included which uses the default timings provided in the function (otherwise, passing a numpy array or list of times to timeWindow will use those times). Optionally, a minimum gamma-ray energy cut-off can also be included, Eth.", "_____no_output_____" ] ], [ [ "times,nubargTime = histTime.nubarg(timeWindow=True) # include timeWindow as a boolean or list of times (in seconds) to activate this feature\nfig = plt.figure(figsize=(8,6))\nplt.plot(times,nubargTime,'o',label='Eth=0. MeV')\ntimes,nubargTime = histTime.nubarg(timeWindow=True,Eth=0.1)\nplt.plot(times,nubargTime,'o',label='Eth=0.1 MeV')\nplt.xlabel('Time since fission (s)')\nplt.ylabel(r'Averge $\\gamma$-ray multiplicity')\nplt.xscale('log')\nplt.legend()\nplt.show()", "_____no_output_____" ] ], [ [ "The prompt fission gamma-ray spectrum function, pfgs(), can also be used to calculate this quantity within a certain time window since the fission event. The time window is defined using minTime and maxTime to set the lower and upper boundaries.", "_____no_output_____" ] ], [ [ "fig = plt.figure(figsize=(8,6))\nbE,pfgsTest = histTime.pfgs(minTime=5e-8,maxTime=500e-8)\nplt.step(bE,pfgsTest,label='Time window')\nbE,pfgsTest = histTime.pfgs()\nplt.step(bE,pfgsTest,label='All events')\nplt.yscale('log')\nplt.xlim(0,2)\nplt.ylim(0.1,100)\nplt.xlabel('Gamma-ray Energy (MeV)')\nplt.ylabel('Prompt Fission Gamma Spectrum')\nplt.legend()\nplt.show()", "_____no_output_____" ], [ "# calculate the gamma-ray multiplicity as a function of time since fission for a specific fission fragment\ntimes,gMultiplicity = histTime.gammaMultiplicity(minTime=1e-8,maxTime=1e-6,Afragment=134,Zfragment=52)\n\n# also compare to an exponential decay with the half life of the state\nf = np.exp(-times*np.log(2)/1.641e-7) # the half life of 134Te is 164.1 ns\nnorm = gMultiplicity[0]/f[0]\n\nfig = plt.figure(figsize=(8,6))\nplt.plot(times*1e9,gMultiplicity/norm,'k-',label='CGMF')\nplt.plot(times*1e9,f,'r--',label=r'exp($-t\\cdot$log(2)/$\\tau_{1/2}$)')\nplt.legend()\nplt.yscale('log')\nplt.xlabel('Time since fission (ns)')\nplt.ylabel(r'N$_\\gamma$(t) (arb. units)')\nplt.show()", "_____no_output_____" ], [ "# calculate the isomeric ratios for specific states in nuclei\n# e.g. isomeric ratio for the 1/2- state in 99Nb, ground state is 9/2+, lifetime is 150 s\nr = histTime.isomericRatio(thresholdTime=1,A=99,Z=41,Jm=0.5,Jgs=4.5)\nprint ('99Nb:',round(r,2)) \n# e.g. isomeric ratio for the 11/2- state in 133Te, ground state is 3/2+, lifetime is 917.4 s\nr = histTime.isomericRatio(thresholdTime=1,A=133,Z=52,Jm=5.5,Jgs=1.5)\nprint ('133Te:',round(r,2))", "99Nb: 1.0\n133Te: 0.14\n" ] ], [ [ "### 9. Angular Correlations\n<a id='angles'></a>", "_____no_output_____" ], [ "For the fission fragment angular distribution with respect to the beam axis/z-axis, there is one option: afterEmission=True/False. afterEmission=True uses these angles after neutron emission and afterEmission=False uses these angles before neutron emission. The default is True.", "_____no_output_____" ] ], [ [ "# calculate cos(theta) between the fragments and z-axis/beam axis\nFFangles = hist.FFangles()\nbins = np.linspace(-1,1,30)\nh,b = np.histogram(FFangles,bins=bins,density=True)\n# only light fragments\nhLight,b = np.histogram(FFangles[::2],bins=bins,density=True)\n# only heavy fragments\nhHeavy,b = np.histogram(FFangles[1::2],bins=bins,density=True)", "_____no_output_____" ], [ "x = 0.5*(b[:-1]+b[1:])\nfig = plt.figure(figsize=(8,6))\nplt.plot(x,h,'k*',label='All Fragments')\nplt.plot(x,hLight,'ro',label='Light Fragments')\nplt.plot(x,hHeavy,'b^',label='Heavy Fragments')\nplt.xlabel(r'cos($\\theta$)')\nplt.ylabel('Frequency')\nplt.ylim(0.45,0.55)\nplt.title('Fission fragment angles with respect to beam axis')\nplt.legend()\nplt.show()", "_____no_output_____" ] ], [ [ "There are several options when calculating the angles of the neutrons with respect to the beam axis/z-axis. The first is including a neutron threshold energy with keyword, Eth (given in MeV). We can also calculate these angles in the lab frame (lab=True, default) or in the center of mass frame of the compound system (lab=False). Finally, we can include pre-fission neutrons (includePrefission=True, default) or not include them (includePreFission=False). However, the pre-fission neutrons can only be include in the lab frame.", "_____no_output_____" ] ], [ [ "# calculate the angles between the neutrons and the z-axis/beam axis\nnAllLab,nLLab,nHLab = hist.nangles(lab=True) # all neutrons, from the light fragment, from the heavy fragment\nnAllCM,nLCM,nHCM = hist.nangles(lab=False) # center of mass frame of the compound", "_____no_output_____" ], [ "bins = np.linspace(-1,1,30)\nhAllLab,b = np.histogram(nAllLab,bins=bins,density=True)\nhLightLab,b = np.histogram(nLLab,bins=bins,density=True)\nhHeavyLab,b = np.histogram(nHLab,bins=bins,density=True)\nhAllcm,b = np.histogram(nAllCM,bins=bins,density=True)\nhLightcm,b = np.histogram(nLCM,bins=bins,density=True)\nhHeavycm,b = np.histogram(nHCM,bins=bins,density=True)", "_____no_output_____" ], [ "x = 0.5*(b[:-1]+b[1:])\nfig = plt.figure(figsize=(8,6))\nplt.plot(x,hAllLab,'k*',label='All Fragments')\nplt.plot(x,hLightLab,'ro',label='Light Fragments')\nplt.plot(x,hHeavyLab,'b^',label='Heavy Fragments')\nplt.xlabel(r'cos($\\theta$)')\nplt.ylabel('Frequency')\nplt.ylim(0.45,0.55)\nplt.title('Neutron Angles with respect to beam axis in the Lab Frame')\nplt.legend()\nplt.show()", "_____no_output_____" ], [ "fig = plt.figure(figsize=(8,6))\nplt.plot(x,hAllLab,'k*',label='Lab Frame')\nplt.plot(x,hAllcm,'ro',label='CoM Frame')\nplt.xlabel(r'cos($\\theta$)')\nplt.ylabel('Frequency')\nplt.ylim(0.45,0.55)\nplt.legend()\nplt.show()", "_____no_output_____" ] ], [ [ "There are again several options that we can use when calculating the angles between all pairs of neutrons (from all framgments) and the ligh fragments, all of which have been seen in the last two examples. These include, Eth (neutron threshold energy), afterEmission (fission fragment angles are post or pre neutron emission), and includePrefission (to include or not include pre-fission neutrons).", "_____no_output_____" ] ], [ [ "# calculate the angles between the neutrons and the light fragments\nnFall,nFLight,nFHeavy = hist.nFangles()\nbins = np.linspace(-1,1,30)\nhall,b = np.histogram(nFall,bins=bins,density=True)\nhlight,b = np.histogram(nFLight,bins=bins,density=True)\nhheavy,b = np.histogram(nFHeavy,bins=bins,density=True)", "_____no_output_____" ], [ "x = 0.5*(b[:-1]+b[1:])\nfig = plt.figure(figsize=(8,6))\nplt.plot(x,hall,'k*',label='All Fragments')\nplt.plot(x,hlight,'ro',label='Light Fragments')\nplt.plot(x,hheavy,'b^',label='Heavy Fragments')\nplt.xlabel(r'cos($\\theta$)')\nplt.ylabel('Frequency')\nplt.legend()\nplt.show()", "_____no_output_____" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "code" ], [ "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# 0. required packages for h5py", "_____no_output_____" ] ], [ [ "%run \"..\\..\\Startup_py3.py\"\nsys.path.append(r\"..\\..\\..\\..\\Documents\")\n\nimport ImageAnalysis3 as ia\n%matplotlib notebook\n\nfrom ImageAnalysis3 import *\nprint(os.getpid())\n\nimport h5py\nfrom ImageAnalysis3.classes import _allowed_kwds\nimport ast", "18112\n" ] ], [ [ "# 1. Create field-of-view class", "_____no_output_____" ] ], [ [ "reload(ia)\nreload(classes)\nreload(classes.batch_functions)\nreload(classes.field_of_view)\nreload(io_tools.load)\n\nreload(visual_tools)\nreload(ia.correction_tools)\nreload(ia.correction_tools.alignment)\nreload(ia.spot_tools.matching)\nreload(ia.segmentation_tools.chromosome)\nreload(ia.spot_tools.fitting)\n\nfov_param = {'data_folder':r'\\\\10.245.74.158\\Chromatin_NAS_1\\20210320-proB_Dox_IAA_STI_CTP-08_2color',\n 'save_folder':r'\\\\10.245.74.212\\Chromatin_NAS_2\\IgH_analyzed_results\\20210320_IgH_proB_iaa_dox+',\n #'save_folder':r'D:\\Pu_Temp\\202009_IgH_proB_DMSO_2color',\n 'experiment_type': 'DNA',\n 'num_threads': 24,\n 'correction_folder':r'\\\\10.245.74.158\\Chromatin_NAS_0\\Corrections\\20201012-Corrections_2color',\n 'shared_parameters':{\n 'single_im_size':[35,2048,2048],\n 'corr_channels':['750','647'],\n 'num_empty_frames': 0, \n 'corr_hot_pixel':True,\n 'corr_Z_shift':False,\n 'min_num_seeds':500,\n 'max_num_seeds': 2500,\n 'spot_seeding_th':125,\n 'normalize_intensity_local':False,\n 'normalize_intensity_background':False,\n }, \n }", "_____no_output_____" ], [ "fov = classes.field_of_view.Field_of_View(fov_param, _fov_id=30,\n _color_info_kwargs={\n '_color_filename':'Color_Usage_clean',\n }, \n _prioritize_saved_attrs=False,\n )", "Get Folder Names: (ia.get_img_info.get_folders)\n- Number of folders: 78\n- Number of field of views: 64\n- Importing csv file: \\\\10.245.74.158\\Chromatin_NAS_1\\20210320-proB_Dox_IAA_STI_CTP-08_2color\\Analysis\\Color_Usage_clean.csv\n- header: ['Hyb', '750', '647', '488', '405']\n-- Hyb H0R0 exists in this data\n-- DAPI exists in hyb: H0R0\n- 75 folders are found according to color-usage annotation.\n+ loading fov_info from file: \\\\10.245.74.212\\Chromatin_NAS_2\\IgH_analyzed_results\\20210320_IgH_proB_iaa_dox+\\Conv_zscan_30.hdf5\n++ base attributes loaded:['be8ce21d23df451e85292b57a8b61273', 'cand_chrom_coords', 'chrom_coords', 'chrom_im', 'ref_im'] in 11.353s.\n+ loading correction from file: \\\\10.245.74.212\\Chromatin_NAS_2\\IgH_analyzed_results\\20210320_IgH_proB_iaa_dox+\\Conv_zscan_30.hdf5\n++ load bleed correction profile directly from savefile.\n++ load chromatic correction profile directly from savefile.\n++ load chromatic_constants correction profile directly from savefile.\n++ load illumination correction profile directly from savefile.\n+ loading segmentation from file: \\\\10.245.74.212\\Chromatin_NAS_2\\IgH_analyzed_results\\20210320_IgH_proB_iaa_dox+\\Conv_zscan_30.hdf5\n++ base attributes loaded:[] in 0.005s.\n-- saving fov_info to file: \\\\10.245.74.212\\Chromatin_NAS_2\\IgH_analyzed_results\\20210320_IgH_proB_iaa_dox+\\Conv_zscan_30.hdf5\n++ base attributes saved:['analysis_folder', 'annotated_folders', 'be8ce21d23df451e85292b57a8b61273', 'bead_channel_index', 'cand_chrom_coords', 'channels', 'chrom_coords', 'chrom_im', 'color_dic', 'color_filename', 'color_format', 'correction_folder', 'dapi_channel_index', 'data_folder', 'drift', 'drift_filename', 'drift_folder', 'experiment_folder', 'folders', 'fov_id', 'fov_name', 'map_folder', 'num_threads', 'ref_filename', 'ref_id', 'ref_im', 'rotation', 'save_filename', 'save_folder', 'segmentation_dim', 'segmentation_folder', 'shared_parameters', 'use_dapi'] in 18.835s.\n" ] ], [ [ "### 2. Process image into candidate spots", "_____no_output_____" ] ], [ [ "reload(io_tools.load)\nreload(spot_tools.fitting)\nreload(correction_tools.chromatic)\nreload(classes.batch_functions)\n\n# process image into spots\nid_list, spot_list = fov._process_image_to_spots('unique', \n #_sel_ids=np.arange(41,47),\n _load_common_reference=True,\n _load_with_multiple=False,\n _save_images=True,\n _warp_images=False, \n _overwrite_drift=False,\n _overwrite_image=False,\n _overwrite_spot=False,\n _verbose=True)", "-- No folder selected, allow processing all 75 folders\n+ load reference image from file:\\\\10.245.74.158\\Chromatin_NAS_1\\20210320-proB_Dox_IAA_STI_CTP-08_2color\\H0R0\\Conv_zscan_30.dax\n- correct the whole fov for image: \\\\10.245.74.158\\Chromatin_NAS_1\\20210320-proB_Dox_IAA_STI_CTP-08_2color\\H0R0\\Conv_zscan_30.dax\n-- loading illumination correction profile from file:\n\t 488 illumination_correction_488_2048x2048.npy\n-- loading image from file:\\\\10.245.74.158\\Chromatin_NAS_1\\20210320-proB_Dox_IAA_STI_CTP-08_2color\\H0R0\\Conv_zscan_30.dax in 8.178s\n-- removing hot pixels for channels:['488'] in 6.360s\n-- illumination correction for channels: 488, in 1.390s\n-- -- generate translation function with drift:[0. 0. 0.] in 0.000s\n-- finish correction in 16.463s\n-- saving fov_info to file: \\\\10.245.74.212\\Chromatin_NAS_2\\IgH_analyzed_results\\20210320_IgH_proB_iaa_dox+\\Conv_zscan_30.hdf5\n++ base attributes saved:['ref_im'] in 4.984s.\n-- checking unique, region:[41 42] in 0.047s.\n-- checking unique, region:[44 45] in 0.000s.\n-- checking unique, region:[47 48] in 0.016s.\n-- checking unique, region:[50 51] in 0.007s.\n-- checking unique, region:[53 54] in 0.000s.\n-- checking unique, region:[56 57] in 0.009s.\n-- checking unique, region:[60 61] in 0.000s.\n-- checking unique, region:[63 64] in 0.000s.\n-- checking unique, region:[66 67] in 0.016s.\n-- checking unique, region:[69 70] in 0.000s.\n-- checking unique, region:[72 73] in 0.000s.\n-- checking unique, region:[75 76] in 0.016s.\n-- checking unique, region:[78 79] in 0.000s.\n-- checking unique, region:[81 82] in 0.000s.\n-- checking unique, region:[84 85] in 0.016s.\n-- checking unique, region:[87 88] in 0.000s.\n-- checking unique, region:[90 91] in 0.000s.\n-- checking unique, region:[93 94] in 0.016s.\n-- checking unique, region:[96 97] in 0.000s.\n-- checking unique, region:[ 99 100] in 0.000s.\n-- checking unique, region:[102 103] in 0.016s.\n-- checking unique, region:[105 106] in 0.000s.\n-- checking unique, region:[108 109] in 0.000s.\n-- checking unique, region:[111 112] in 0.016s.\n-- checking unique, region:[114 115] in 0.000s.\n-- checking unique, region:[323 321] in 0.000s.\n-- checking unique, region:[326 324] in 0.016s.\n-- checking unique, region:[329 327] in 0.000s.\n-- checking unique, region:[332 330] in 0.000s.\n-- checking unique, region:[335 333] in 0.016s.\n-- checking unique, region:[339 337] in 0.000s.\n-- checking unique, region:[342 340] in 0.000s.\n-- checking unique, region:[345 343] in 0.016s.\n-- checking unique, region:[348 346] in 0.000s.\n-- checking unique, region:[351 349] in 0.000s.\n-- checking unique, region:[354 352] in 0.016s.\n-- checking unique, region:[357 355] in 0.000s.\n-- checking unique, region:[360 358] in 0.000s.\n-- checking unique, region:[363 361] in 0.017s.\n-- checking unique, region:[366 364] in 0.001s.\n-- checking unique, region:[369 367] in 0.000s.\n-- checking unique, region:[375 373] in 0.014s.\n-- checking unique, region:[388 383] in 0.000s.\n-- checking unique, region:[391 386] in 0.013s.\n-- checking unique, region:[394 389] in 0.006s.\n-- checking unique, region:[ 43 392] in 0.006s.\n-- checking unique, region:[ 49 395] in 0.005s.\n-- checking unique, region:[55 46] in 0.006s.\n-- checking unique, region:[62 52] in 0.005s.\n-- checking unique, region:[68 59] in 0.005s.\n-- checking unique, region:[74 65] in 0.004s.\n-- checking unique, region:[80 71] in 0.005s.\n-- checking unique, region:[86 77] in 0.005s.\n-- checking unique, region:[92 83] in 0.005s.\n-- checking unique, region:[98 89] in 0.005s.\n-- checking unique, region:[104 95] in 0.005s.\n-- checking unique, region:[110 101] in 0.004s.\n-- checking unique, region:[325 107] in 0.000s.\n-- checking unique, region:[331 113] in 0.000s.\n-- checking unique, region:[341 328] in 0.014s.\n-- checking unique, region:[347 334] in 0.000s.\n-- checking unique, region:[353 344] in 0.000s.\n-- checking unique, region:[359 350] in 0.016s.\n-- checking unique, region:[365 356] in 0.000s.\n-- checking unique, region:[371 362] in 0.000s.\n-- checking unique, region:[377 368] in 0.016s.\n-- checking unique, region:[384 374] in 0.000s.\n-- checking unique, region:[390 381] in 0.000s.\n-- checking unique, region:[393 387] in 0.017s.\n+ Start multi-processing of pre-processing for 69 images with 24 threads\n++ processed unique ids: [ 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 59\n 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77\n 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95\n 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113\n 114 115 321 323 324 325 326 327 328 329 330 331 332 333 334 335 337 339\n 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357\n 358 359 360 361 362 363 364 365 366 367 368 369 371 373 374 375 377 381\n 383 384 386 387 388 389 390 391 392 393 394 395] in 3189.44s.\n" ] ], [ [ "# 3. Find chromosomes", "_____no_output_____" ], [ "## 3.1 load chromosome image", "_____no_output_____" ] ], [ [ "overwrite_chrom = False\nchrom_im = fov._load_chromosome_image(_type='reverse', \n _overwrite=overwrite_chrom)", "-- choose chrom images from folder: \\.\n- correct the whole fov for image: \\\\10.245.74.158\\Chromatin_NAS_1\\20210320-proB_Dox_IAA_STI_CTP-08_2color\\H0R0\\Conv_zscan_30.dax\n-- loading illumination correction profile from file:\n\t 647 illumination_correction_647_2048x2048.npy\n-- loading chromatic correction profile from file:\n\t 750 chromatic_correction_750_647_35_2048_2048.npy\n\t 647 None\n-- loading image from file:\\\\10.245.74.158\\Chromatin_NAS_1\\20210320-proB_Dox_IAA_STI_CTP-08_2color\\H0R0\\Conv_zscan_30.dax in 8.030s\n-- removing hot pixels for channels:['647'] in 7.033s\n-- illumination correction for channels: 647, in 1.573s\n-- warp image with chromatic correction for channels: [] and drift:[0. 0. 0.] in 0.000s\n-- finish correction in 38.839s\n-- chromosome image has drift: [0. 0. 0.]\n-- saving fov_info to file: \\\\10.245.74.212\\Chromatin_NAS_2\\IgH_analyzed_results\\20210320_IgH_proB_iaa_dox+\\Conv_zscan_30.hdf5\n++ base attributes saved:['chrom_im'] in 5.277s.\n" ] ], [ [ "## 3.2 find candidate chromosomes", "_____no_output_____" ] ], [ [ "chrom_coords = fov._find_candidate_chromosomes_by_segmentation(_filt_size=4,\n _binary_per_th=99.75, \n _morphology_size=2,\n _overwrite=overwrite_chrom)", "+ loading fov_info from file: \\\\10.245.74.212\\Chromatin_NAS_2\\IgH_analyzed_results\\20210320_IgH_proB_iaa_dox+\\Conv_zscan_30.hdf5\n++ base attributes loaded:[] in 4.192s.\n-- adjust seed image with filter size=4\n-- binarize image with threshold: 99.75%\n-- erosion and dialation with size=2.\n-- find close objects.\n-- random walk segmentation, beta=10.\n" ] ], [ [ "## 3.3 select among candidate chromosomes", "_____no_output_____" ] ], [ [ "chrom_coords = fov._select_chromosome_by_candidate_spots(_good_chr_loss_th=0.3,\n _cand_spot_intensity_th=200,\n _save=True, \n _overwrite=overwrite_chrom)", "+ loading fov_info from file: \\\\10.245.74.212\\Chromatin_NAS_2\\IgH_analyzed_results\\20210320_IgH_proB_iaa_dox+\\Conv_zscan_30.hdf5\n++ base attributes loaded:[] in 5.189s.\n+ loading unique from file: \\\\10.245.74.212\\Chromatin_NAS_2\\IgH_analyzed_results\\20210320_IgH_proB_iaa_dox+\\Conv_zscan_30.hdf5\n" ] ], [ [ "### visualize chromosomes selections", "_____no_output_____" ] ], [ [ "%matplotlib notebook\n%matplotlib notebook\n## visualize\ncoord_dict = {'coords':[np.flipud(_coord) for _coord in fov.chrom_coords],\n 'class_ids':list(np.zeros(len(fov.chrom_coords),dtype=np.int)),\n }\n\nvisual_tools.imshow_mark_3d_v2([fov.chrom_im], \n given_dic=coord_dict,\n save_file=None,\n )\n", "_____no_output_____" ] ], [ [ "## select spots based on chromosomes", "_____no_output_____" ] ], [ [ "fov._load_from_file('unique')", "+ loading unique from file: \\\\10.245.74.212\\Chromatin_NAS_2\\IgH_analyzed_results\\20210320_IgH_proB_iaa_dox+\\Conv_zscan_30.hdf5\n++ finish loading unique in 0.836s. \n" ], [ "intensity_th = 200\nfrom ImageAnalysis3.spot_tools.picking import assign_spots_to_chromosomes\n\nkept_spots_list = []\nfor _spots in fov.unique_spots_list:\n kept_spots_list.append(_spots[_spots[:,0] > intensity_th])\n# finalize candidate spots\ncand_chr_spots_list = [[] for _ct in fov.chrom_coords]\nfor _spots in kept_spots_list:\n _cands_list = assign_spots_to_chromosomes(_spots, fov.chrom_coords)\n for _i, _cands in enumerate(_cands_list):\n cand_chr_spots_list[_i].append(_cands)\nprint(f\"kept chromosomes: {len(fov.chrom_coords)}\")", "kept chromosomes: 549\n" ], [ "reload(spot_tools.picking)\nfrom ImageAnalysis3.spot_tools.picking import convert_spots_to_hzxys\n\ndna_cand_hzxys_list = [convert_spots_to_hzxys(_spots, fov.shared_parameters['distance_zxy'])\n for _spots in cand_chr_spots_list]\ndna_reg_ids = fov.unique_ids\ndna_reg_channels = fov.unique_channels\nchrom_coords = fov.chrom_coords\n\n\n# select_hzxys close to the chromosome center\ndist_th = 3000 # upper limit is 3000nm\ngood_chr_th = 0.8 # 80% of regions should have candidate spots\n\nsel_dna_cand_hzxys_list = []\nsel_chrom_coords = []\nchr_cand_pers = []\nsel_chr_cand_pers = []\nfor _cand_hzxys, _chrom_coord in zip(dna_cand_hzxys_list, chrom_coords):\n _chr_cand_per = 0\n _sel_cands_list = []\n \n for _cands in _cand_hzxys:\n if len(_cands) == 0:\n _sel_cands_list.append([])\n else:\n _dists = np.linalg.norm(_cands[:,1:4] - _chrom_coord*np.array([200,108,108]), axis=1)\n _sel_cands_list.append(_cands[(_dists < dist_th)])\n _chr_cand_per += 1\n \n _chr_cand_per *= 1/len(_cand_hzxys)\n # append\n if _chr_cand_per >= good_chr_th:\n sel_dna_cand_hzxys_list.append(_sel_cands_list)\n sel_chrom_coords.append(_chrom_coord)\n sel_chr_cand_pers.append(_chr_cand_per)\n \n chr_cand_pers.append(_chr_cand_per)\n \nprint(f\"kept chromosomes: {len(sel_chrom_coords)}\")", "kept chromosomes: 522\n" ] ], [ [ "### EM pick spots", "_____no_output_____" ] ], [ [ "%matplotlib inline\nreload(spot_tools.picking)\nfrom ImageAnalysis3.spot_tools.picking import _maximize_score_spot_picking_of_chr, pick_spots_by_intensities,pick_spots_by_scores, generate_reference_from_population, evaluate_differences\n\nniter= 10\nnum_threads = 24\nref_chr_cts = None\n# initialize\ninit_dna_hzxys = pick_spots_by_intensities(sel_dna_cand_hzxys_list)\n# set save list\nsel_dna_hzxys_list, sel_dna_scores_list, all_dna_scores_list = [init_dna_hzxys], [], []\n\nfor _iter in range(niter):\n print(f\"+ iter:{_iter}\")\n # E: generate reference\n ref_ct_dists, ref_local_dists, ref_ints = generate_reference_from_population(\n sel_dna_hzxys_list[-1], dna_reg_ids, \n sel_dna_hzxys_list[-1], dna_reg_ids,\n ref_channels=dna_reg_channels,\n ref_chr_cts=ref_chr_cts,\n num_threads=num_threads,\n collapse_regions=True,\n split_channels=True,\n verbose=True,\n )\n \n plt.figure(figsize=(4,2), dpi=100)\n for _k, _v in ref_ct_dists.items():\n plt.hist(np.array(_v), bins=np.arange(0,2500,50), alpha=0.5, label=_k)\n plt.legend(fontsize=8)\n plt.title('center dist', fontsize=8)\n plt.show()\n \n plt.figure(figsize=(4,2), dpi=100)\n for _k, _v in ref_local_dists.items():\n plt.hist(np.array(_v), bins=np.arange(0,2500,50), alpha=0.5, label=_k)\n plt.legend(fontsize=8)\n plt.title('local dist', fontsize=8)\n plt.show()\n \n plt.figure(figsize=(4,2), dpi=100)\n for _k, _v in ref_ints.items():\n plt.hist(np.array(_v), bins=np.arange(0,5000,100), alpha=0.5, label=_k)\n plt.legend(fontsize=8)\n plt.title('intensity', fontsize=8)\n plt.show()\n \n # M: pick based on scores\n sel_hzxys_list, sel_scores_list, all_scores_list, other_scores_list = \\\n pick_spots_by_scores(\n sel_dna_cand_hzxys_list, dna_reg_ids,\n ref_channels=dna_reg_channels,\n ref_hzxys_list=sel_dna_hzxys_list[-1], ref_ids=dna_reg_ids, \n ref_ct_dists=ref_ct_dists, ref_local_dists=ref_local_dists, ref_ints=ref_ints, \n ref_chr_cts=ref_chr_cts,\n num_threads=num_threads,\n collapse_regions=True,\n split_intensity_channels=True,\n split_distance_channels=False,\n return_other_scores=True,\n verbose=True,\n )\n # check updating rate\n update_rate = evaluate_differences(sel_hzxys_list, sel_dna_hzxys_list[-1])\n print(f\"-- region kept: {update_rate:.4f}\")\n # append\n sel_dna_hzxys_list.append(sel_hzxys_list)\n sel_dna_scores_list.append(sel_scores_list)\n all_dna_scores_list.append(all_scores_list)\n \n plt.figure(figsize=(4,2), dpi=100)\n plt.hist(np.concatenate([np.concatenate(_scores) \n for _scores in other_scores_list]), \n bins=np.arange(-15, 0, 0.5), alpha=0.5, label='unselected')\n plt.hist(np.ravel([np.array(_sel_scores) \n for _sel_scores in sel_dna_scores_list[-1]]), \n bins=np.arange(-15, 0, 0.5), alpha=0.5, label='selected')\n plt.legend(fontsize=8)\n plt.show()\n \n if update_rate > 0.998:\n break", "+ iter:0\n-- generate reference metrics\n--- multiprocessing expectation step with 24 threads, in 12.560s\n--- collapse all regions into 1d.\n" ], [ "%%timeit \nspot_tools.picking.chromosome_center_dists(sel_dna_hzxys_list[0][0], ref_channels=dna_reg_channels, split_channels=False)", "1.57 ms ± 13.6 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n" ], [ "%%timeit \nspot_tools.picking.chromosome_center_dists(sel_dna_hzxys_list[0][0], ref_channels=dna_reg_channels, split_channels=True)", "1.78 ms ± 10.7 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n" ], [ "from scipy.spatial.distance import pdist, squareform\nsel_iter = -1\n\nfinal_dna_hzxys_list = []\nkept_chr_ids = []\ndistmap_list = []\nscore_th = -5\nint_th = 200\nbad_spot_percentage = 1.0 #0.5\n\nfor _hzxys, _scores in zip(sel_dna_hzxys_list[sel_iter], sel_dna_scores_list[sel_iter]):\n _kept_hzxys = np.array(_hzxys).copy()\n # remove spots by intensity\n _bad_inds = _kept_hzxys[:,0] < int_th\n # remove spots by scores\n _bad_inds += _scores < score_th\n #print(np.mean(_bad_inds))\n _kept_hzxys[_bad_inds] = np.nan\n \n \n if np.mean(np.isnan(_kept_hzxys).sum(1)>0)<bad_spot_percentage:\n kept_chr_ids.append(True)\n final_dna_hzxys_list.append(_kept_hzxys)\n distmap_list.append(squareform(pdist(_kept_hzxys[:,1:4])))\n else:\n kept_chr_ids.append(False)\n\nkept_chr_ids = np.array(kept_chr_ids, dtype=np.bool)\n#kept_chrom_coords = np.array(sel_chrom_coords)[kept_chr_ids]\ndistmap_list = np.array(distmap_list)\nmedian_distmap = np.nanmedian(distmap_list, axis=0)", "E:\\Users\\puzheng\\anaconda3\\lib\\site-packages\\numpy\\lib\\nanfunctions.py:1114: RuntimeWarning: All-NaN slice encountered\n overwrite_input=overwrite_input)\n" ], [ "loss_rates = np.mean(np.sum(np.isnan(final_dna_hzxys_list), axis=2)>0, axis=0)\nprint(np.mean(loss_rates))\nfig, ax = plt.subplots(figsize=(4,2),dpi=200)\nax.plot(loss_rates, '.-')\nax.set_ylim([0,1])\nax.set_xticks(np.arange(0,len(dna_reg_ids),int(len(dna_reg_ids)/5)))\nplt.show()", "0.26291021156088623\n" ], [ "imaging_order = []\nfor _fd, _infos in fov.color_dic.items():\n for _info in _infos:\n if len(_info) > 0 and _info[0] == 'u':\n if int(_info[1:]) in dna_reg_ids:\n imaging_order.append(list(dna_reg_ids).index(int(_info[1:])))\nimaging_order = np.array(imaging_order, dtype=np.int)\n#kept_inds = imaging_order # plot imaging ordered regions\n\n#kept_inds = np.where(loss_rates<0.5)[0] # plot good regions only\nkept_inds = np.arange(len(fov.unique_ids)) # plot all\n\n%matplotlib inline\n\nfig, ax = plt.subplots(figsize=(4,3),dpi=200)\nax = ia.figure_tools.distmap.plot_distance_map(median_distmap[kept_inds][:,kept_inds], \n color_limits=[0,600],\n ax=ax,\n ticks=np.arange(0,150,20), \n figure_dpi=500)\nax.set_title(f\"v-Abl ProB iaa_dox_STI+, n={len(distmap_list)}\", fontsize=7.5)\n\n_ticks = np.arange(0, len(kept_inds), 20)\nax.set_xticks(_ticks)\nax.set_xticklabels(dna_reg_ids[kept_inds][_ticks])\nax.set_xlabel(f\"5kb region id\", fontsize=7, labelpad=2)\nax.set_yticks(_ticks)\nax.set_yticklabels(dna_reg_ids[kept_inds][_ticks])\nax.set_ylabel(f\"5kb region id\", fontsize=7, labelpad=2)\n\n\nax.axvline(x=np.where(fov.unique_ids[kept_inds]>300)[0][0], color=[1,1,0])\nax.axhline(y=np.where(fov.unique_ids[kept_inds]>300)[0][0], color=[1,1,0])\n\nplt.gcf().subplots_adjust(bottom=0.1)\nplt.show()", "_____no_output_____" ] ], [ [ "## visualize single example", "_____no_output_____" ] ], [ [ "%matplotlib inline\n\n\nreload(figure_tools.image)\n\nchrom_id = 3\n\nimport matplotlib\nimport copy\n\nsc_cmap = copy.copy(matplotlib.cm.get_cmap('seismic_r'))\nsc_cmap.set_bad(color=[0.5,0.5,0.5,1])\n\n#valid_inds = np.where(np.isnan(final_dna_hzxys_list[chrom_id]).sum(1) == 0)[0]\nvalid_inds = np.ones(len(final_dna_hzxys_list[chrom_id]), dtype=np.bool) # all spots\n\nfig, ax = plt.subplots(figsize=(4,3),dpi=200)\nax = ia.figure_tools.distmap.plot_distance_map(\n distmap_list[chrom_id][valid_inds][:,valid_inds], \n color_limits=[0,600],\n ax=ax,\n cmap=sc_cmap,\n ticks=np.arange(0,150,20), \n figure_dpi=200)\nax.set_title(f\"proB DMSO chrom: {chrom_id}\", fontsize=7.5)\nplt.gcf().subplots_adjust(bottom=0.1)\nplt.show()\n\nax3d = figure_tools.image.chromosome_structure_3d_rendering(\n final_dna_hzxys_list[chrom_id][valid_inds, 1:], \n marker_edge_line_width=0,\n reference_bar_length=200, image_radius=300, \n line_width=0.5, figure_dpi=300, depthshade=False)\nplt.show()", "_____no_output_____" ], [ "?figure_tools.image.chromosome_structure_3d_rendering", "_____no_output_____" ] ], [ [ "## visualize all fitted spots", "_____no_output_____" ] ], [ [ "with h5py.File(fov.save_filename, \"r\", libver='latest') as _f:\n _grp = _f['unique']\n raw_spots_list = [_spots[_spots[:,0] > 0] for _spots in _grp['raw_spots'][:]]\n spots_list = [_spots[_spots[:,0] > 0] for _spots in _grp['spots'][:]]", "_____no_output_____" ], [ "from scipy.spatial.distance import cdist\npicked_spot_inds_list = []\nfor _i, _id in enumerate(dna_reg_ids):\n _cand_hzxys = spots_list[_i][:,1:4] * fov.shared_parameters['distance_zxy']\n _dists = cdist(np.array(final_dna_hzxys_list)[:,_i,1:], _cand_hzxys)#, axis=1\n _matched_spot_inds = []\n for _ds in _dists:\n if np.sum(np.isnan(_ds)) < len(_ds) and np.nanmin(_ds) < 0.01:\n _matched_spot_inds.append(np.argmin(_ds))\n else:\n _matched_spot_inds.append(np.nan)\n # append\n picked_spot_inds_list.append(np.array(_matched_spot_inds))", "_____no_output_____" ], [ "#vis_inds = [0,1,2,3,4,5]\nvis_inds = np.where(loss_rates > 0.8)[0]\nvis_ims, vis_ids, vis_spot_list, vis_raw_spot_list = [], [], [], []\n\nwith h5py.File(fov.save_filename, \"r\", libver='latest') as _f:\n _grp = _f['unique']\n \n for _ind in vis_inds:\n vis_ims.append(_grp['ims'][_ind])\n vis_ids.append(_grp['ids'][_ind])\n _picked_inds = picked_spot_inds_list[_ind]\n _picked_inds = np.array(_picked_inds[np.isnan(_picked_inds)==False], dtype=np.int)\n vis_spot_list.append(raw_spots_list[_ind][_picked_inds])", "_____no_output_____" ], [ "dna_reg_ids[59]", "_____no_output_____" ], [ "fov.color_dic", "_____no_output_____" ], [ "# visualize_all_chromosomes\n%matplotlib notebook\n%matplotlib notebook\n\n## visualize\ncoord_dict = {'coords':[],\n 'class_ids':[],\n }\nfor _i, _spots in enumerate(vis_spot_list):\n coord_dict['coords'] += list(np.flipud(_spot[1:4]) for _spot in _spots)\n coord_dict['class_ids'] += list(_i * np.ones(len(_spots),dtype=np.int))\n\nfig=plt.figure(figsize=(4,6), dpi=150) \n \nvisual_tools.imshow_mark_3d_v2(vis_ims, \n fig=fig,\n given_dic=coord_dict,\n save_file=None,\n )", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "CIFAR10 是另外一個 dataset， 和 mnist 一樣，有十種類別（飛機、汽車、鳥、貓、鹿、狗、青蛙、馬、船、卡車）\n\nhttps://www.cs.toronto.edu/~kriz/cifar.html\n", "_____no_output_____" ] ], [ [ "import keras\nfrom keras.models import Sequential\nfrom PIL import Image\nimport numpy as np\nimport tarfile\n", "Using TensorFlow backend.\n" ], [ "# 讀取 dataset\n# 只有 train 和 test 沒有 validation\nimport pickle\ntrain_X=[]\ntrain_y=[]\ntar_gz = \"../Week06/cifar-10-python.tar.gz\"\nwith tarfile.open(tar_gz) as tarf:\n for i in range(1, 6):\n dataset = \"cifar-10-batches-py/data_batch_%d\"%i\n print(\"load\",dataset)\n with tarf.extractfile(dataset) as f:\n result = pickle.load(f, encoding='latin1')\n train_X.extend(result['data']/255)\n train_y.extend(result['labels'])\n train_X=np.float32(train_X)\n train_y=np.int32(train_y)\n dataset = \"cifar-10-batches-py/test_batch\"\n print(\"load\",dataset)\n with tarf.extractfile(dataset) as f:\n result = pickle.load(f, encoding='latin1')\n test_X=np.float32(result['data']/255)\n test_y=np.int32(result['labels'])\ntrain_Y = np.eye(10)[train_y]\ntest_Y = np.eye(10)[test_y]\nvalidation_data = (test_X[:1000], test_Y[:1000])\ntest_data = (test_X[1000:], test_Y[1000:])\n", "load cifar-10-batches-py/data_batch_1\nload cifar-10-batches-py/data_batch_2\nload cifar-10-batches-py/data_batch_3\nload cifar-10-batches-py/data_batch_4\nload cifar-10-batches-py/data_batch_5\nload cifar-10-batches-py/test_batch\n" ], [ "from IPython.display import display\ndef showX(X):\n int_X = (X*255).clip(0,255).astype('uint8')\n # N*3072 -> N*3*32*32 -> 32 * 32N * 3\n int_X_reshape = np.moveaxis(int_X.reshape(-1,3,32,32), 1, 3)\n int_X_reshape = int_X_reshape.swapaxes(0,1).reshape(32,-1, 3)\n display(Image.fromarray(int_X_reshape))\n# 訓練資料， X 的前 20 筆\nshowX(train_X[:20])\nprint(train_y[:20])\nname_array = np.array(\"飛機、汽車、鳥、貓、鹿、狗、青蛙、馬、船、卡車\".split('、'))\nprint(name_array[train_y[:20]])", "_____no_output_____" ] ], [ [ "將之前的 cnn model 套用過來看看", "_____no_output_____" ] ], [ [ "# %load ../Week06/q_cifar10_cnn.py\nimport keras\nfrom keras.layers import Dense, Activation, Conv2D, MaxPool2D, Reshape\nmodel = Sequential()\nmodel.add(Reshape((3, 32, 32), input_shape=(3*32*32,) ))\nmodel.add(Conv2D(filters=32, kernel_size=(3,3), padding='same', activation=\"relu\", data_format='channels_first'))\nmodel.add(MaxPool2D())\nmodel.add(Conv2D(filters=64, kernel_size=(3,3), padding='same', activation=\"relu\", data_format='channels_first'))\nmodel.add(MaxPool2D())\nmodel.add(Reshape((-1,)))\nmodel.add(Dense(units=1024, activation=\"relu\"))\nmodel.add(Dense(units=10, activation=\"softmax\"))\nmodel.compile(loss='categorical_crossentropy',\n optimizer='adam',\n metrics=['accuracy'])\nmodel.fit(train_X, train_Y, validation_data=validation_data, batch_size=100, epochs=10)\nrtn = model.evaluate(*test_data)\nprint(\"\\ntest accuracy=\", rtn[1])", "Train on 50000 samples, validate on 1000 samples\nEpoch 1/10\n50000/50000 [==============================] - 21s - loss: 1.5073 - acc: 0.4609 - val_loss: 1.2153 - val_acc: 0.5720\nEpoch 2/10\n50000/50000 [==============================] - 19s - loss: 1.0979 - acc: 0.6102 - val_loss: 1.0003 - val_acc: 0.6350\nEpoch 3/10\n50000/50000 [==============================] - 19s - loss: 0.9107 - acc: 0.6794 - val_loss: 0.9248 - val_acc: 0.6630\nEpoch 4/10\n50000/50000 [==============================] - 19s - loss: 0.7674 - acc: 0.7314 - val_loss: 0.8943 - val_acc: 0.6900\nEpoch 5/10\n50000/50000 [==============================] - 19s - loss: 0.6320 - acc: 0.7805 - val_loss: 0.8766 - val_acc: 0.6940\nEpoch 6/10\n50000/50000 [==============================] - 19s - loss: 0.4970 - acc: 0.8274 - val_loss: 0.9347 - val_acc: 0.6880\nEpoch 7/10\n50000/50000 [==============================] - 19s - loss: 0.3600 - acc: 0.8763 - val_loss: 0.9964 - val_acc: 0.6850\nEpoch 8/10\n50000/50000 [==============================] - 19s - loss: 0.2404 - acc: 0.9215 - val_loss: 1.0836 - val_acc: 0.6910\nEpoch 9/10\n50000/50000 [==============================] - 21s - loss: 0.1482 - acc: 0.9538 - val_loss: 1.2594 - val_acc: 0.6850\nEpoch 10/10\n50000/50000 [==============================] - 27s - loss: 0.0950 - acc: 0.9724 - val_loss: 1.3658 - val_acc: 0.6980\n8992/9000 [============================>.] - ETA: 0s\ntest accuracy= 0.689333333333\n" ], [ "showX(test_X[:15])\npredict_y = model.predict_classes(test_X[:15], verbose=False)\nprint(name_array[predict_y])\nprint(name_array[test_y[:15]])", "_____no_output_____" ], [ "import keras\nfrom keras.layers import Dense, Activation, Conv2D, MaxPool2D, Reshape, Dropout\nmodel = Sequential()\nmodel.add(Reshape((3, 32, 32), input_shape=(3*32*32,) ))\nmodel.add(Conv2D(32, 3, padding='same', activation=\"relu\", data_format='channels_first'))\nmodel.add(MaxPool2D())\nmodel.add(Conv2D(64, 3, padding='same', activation=\"relu\", data_format='channels_first'))\nmodel.add(MaxPool2D())\nmodel.add(Reshape((-1,)))\nmodel.add(Dense(units=1024, activation=\"relu\"))\nmodel.add(Dropout(rate=0.4))\nmodel.add(Dense(units=10, activation=\"softmax\"))\nmodel.compile(loss='categorical_crossentropy',\n optimizer='adam',\n metrics=['accuracy'])\nmodel.fit(train_X, train_Y, validation_data=validation_data, batch_size=100, epochs=10)\nrtn = model.evaluate(*test_data)\nprint(\"\\ntest accuracy=\", rtn[1])", "Train on 50000 samples, validate on 1000 samples\nEpoch 1/10\n50000/50000 [==============================] - 29s - loss: 1.5101 - acc: 0.4566 - val_loss: 1.2131 - val_acc: 0.5510\nEpoch 2/10\n50000/50000 [==============================] - 28s - loss: 1.1424 - acc: 0.5938 - val_loss: 1.0331 - val_acc: 0.6420\nEpoch 3/10\n50000/50000 [==============================] - 29s - loss: 0.9835 - acc: 0.6518 - val_loss: 0.9500 - val_acc: 0.6670\nEpoch 4/10\n50000/50000 [==============================] - 20s - loss: 0.8547 - acc: 0.6977 - val_loss: 0.8837 - val_acc: 0.6850\nEpoch 5/10\n50000/50000 [==============================] - 19s - loss: 0.7475 - acc: 0.7369 - val_loss: 0.8432 - val_acc: 0.7010\nEpoch 6/10\n50000/50000 [==============================] - 19s - loss: 0.6412 - acc: 0.7723 - val_loss: 0.8798 - val_acc: 0.6880\nEpoch 7/10\n50000/50000 [==============================] - 19s - loss: 0.5418 - acc: 0.8087 - val_loss: 0.8449 - val_acc: 0.7030\nEpoch 8/10\n50000/50000 [==============================] - 19s - loss: 0.4443 - acc: 0.8445 - val_loss: 0.8894 - val_acc: 0.6980\nEpoch 9/10\n50000/50000 [==============================] - 19s - loss: 0.3657 - acc: 0.8729 - val_loss: 0.9427 - val_acc: 0.6860\nEpoch 10/10\n50000/50000 [==============================] - 19s - loss: 0.3048 - acc: 0.8943 - val_loss: 0.9797 - val_acc: 0.6930\n9000/9000 [==============================] - 1s \n\ntest accuracy= 0.702444444444\n" ], [ "model.fit(train_X, train_Y, validation_data=validation_data, batch_size=100, epochs=10)\nrtn = model.evaluate(*test_data)\nprint(\"\\ntest accuracy=\", rtn[1])", "Train on 50000 samples, validate on 1000 samples\nEpoch 1/10\n50000/50000 [==============================] - 19s - loss: 0.2504 - acc: 0.9149 - val_loss: 1.0555 - val_acc: 0.7100\nEpoch 2/10\n50000/50000 [==============================] - 19s - loss: 0.2059 - acc: 0.9283 - val_loss: 1.0203 - val_acc: 0.7090\nEpoch 3/10\n50000/50000 [==============================] - 19s - loss: 0.1726 - acc: 0.9415 - val_loss: 1.1048 - val_acc: 0.7000\nEpoch 4/10\n50000/50000 [==============================] - 19s - loss: 0.1542 - acc: 0.9465 - val_loss: 1.1567 - val_acc: 0.7000\nEpoch 5/10\n50000/50000 [==============================] - 19s - loss: 0.1387 - acc: 0.9524 - val_loss: 1.2201 - val_acc: 0.6980\nEpoch 6/10\n50000/50000 [==============================] - 19s - loss: 0.1261 - acc: 0.9572 - val_loss: 1.2735 - val_acc: 0.7040\nEpoch 7/10\n50000/50000 [==============================] - 19s - loss: 0.1148 - acc: 0.9609 - val_loss: 1.2635 - val_acc: 0.6950\nEpoch 8/10\n50000/50000 [==============================] - 19s - loss: 0.1076 - acc: 0.9632 - val_loss: 1.2785 - val_acc: 0.6860\nEpoch 9/10\n50000/50000 [==============================] - 19s - loss: 0.0964 - acc: 0.9673 - val_loss: 1.4299 - val_acc: 0.6860\nEpoch 10/10\n50000/50000 [==============================] - 19s - loss: 0.0961 - acc: 0.9680 - val_loss: 1.3541 - val_acc: 0.6950\n8736/9000 [============================>.] - ETA: 0s\ntest accuracy= 0.690333333333\n" ], [ "showX(test_X[:15])\npredict_y = model.predict_classes(test_X[:15], verbose=False)\nprint(name_array[predict_y])\nprint(name_array[test_y[:15]])", "_____no_output_____" ] ], [ [ "不同的 activation\nhttps://keras.io/activations/", "_____no_output_____" ] ], [ [ "# 先定義一個工具\ndef add_layers(model, *layers):\n for l in layers:\n model.add(l)", "_____no_output_____" ], [ "import keras\nfrom keras.engine.topology import Layer\nfrom keras.layers import Dense, Activation, Conv2D, MaxPool2D, Reshape, Dropout\n\ndef MyConv2D(filters, kernel_size, **kwargs):\n return (Conv2D(filters=filters, kernel_size=kernel_size, \n padding='same', data_format='channels_first', **kwargs),\n Activation(\"elu\"))\nmodel = Sequential()\nadd_layers( model,\n Reshape((3, 32, 32), input_shape=(3*32*32,)),\n *MyConv2D(32, 3),\n MaxPool2D(),\n *MyConv2D(64, 3),\n MaxPool2D(),\n Reshape((-1,)),\n Dense(units=1024, activation=\"elu\"),\n Dropout(rate=0.4),\n Dense(units=10, activation=\"softmax\")\n)\nmodel.compile(loss='categorical_crossentropy',\n optimizer='adam',\n metrics=['accuracy'])\nmodel.fit(train_X, train_Y, validation_data=validation_data, batch_size=100, epochs=10)\nrtn = model.evaluate(*test_data)\nprint(\"\\ntest accuracy=\", rtn[1])", "Train on 50000 samples, validate on 1000 samples\nEpoch 1/10\n50000/50000 [==============================] - 19s - loss: 1.4263 - acc: 0.5103 - val_loss: 1.0930 - val_acc: 0.6240\nEpoch 2/10\n50000/50000 [==============================] - 19s - loss: 1.0308 - acc: 0.6406 - val_loss: 0.9959 - val_acc: 0.6400\nEpoch 3/10\n50000/50000 [==============================] - 18s - loss: 0.9083 - acc: 0.6827 - val_loss: 0.9100 - val_acc: 0.6790\nEpoch 4/10\n50000/50000 [==============================] - 18s - loss: 0.8233 - acc: 0.7120 - val_loss: 0.9503 - val_acc: 0.6750\nEpoch 5/10\n50000/50000 [==============================] - 19s - loss: 0.7499 - acc: 0.7368 - val_loss: 0.9357 - val_acc: 0.6810\nEpoch 6/10\n50000/50000 [==============================] - 18s - loss: 0.6773 - acc: 0.7627 - val_loss: 0.9786 - val_acc: 0.6790\nEpoch 7/10\n50000/50000 [==============================] - 19s - loss: 0.6140 - acc: 0.7839 - val_loss: 1.0009 - val_acc: 0.6740\nEpoch 8/10\n50000/50000 [==============================] - 19s - loss: 0.5556 - acc: 0.8050 - val_loss: 1.0503 - val_acc: 0.6660\nEpoch 9/10\n50000/50000 [==============================] - 19s - loss: 0.4912 - acc: 0.8256 - val_loss: 1.1419 - val_acc: 0.6610\nEpoch 10/10\n50000/50000 [==============================] - 19s - loss: 0.4269 - acc: 0.8477 - val_loss: 1.1387 - val_acc: 0.6710\n8640/9000 [===========================>..] - ETA: 0s\ntest accuracy= 0.653666666667\n" ], [ "import keras\nfrom keras.layers import Dense, Activation, Conv2D, MaxPool2D, Reshape, Dropout, BatchNormalization\n# 先處理資料\n\n#GCN\ntrain_X_mean = np.mean(train_X, axis=0, keepdims=True)\ntrain_X_std = np.std(train_X, axis=0, keepdims=True)\npreprocessed_train_X = (train_X-train_X_mean)/train_X_std\npreprocessed_test_X = (test_X-train_X_mean)/train_X_std\npreprocessed_validation_data = (preprocessed_test_X[:1000], test_Y[:1000])\npreprocessed_test_data = (preprocessed_test_X[1000:], test_Y[1000:])\n\n\ndef MyConv2D(filters, kernel_size, **kwargs):\n return (Conv2D(filters=filters, kernel_size=kernel_size, \n padding='same', data_format='channels_first', **kwargs),\n Activation(\"relu\"))\ndef add_layers(model, *layers):\n for l in layers:\n model.add(l)\nmodel = Sequential()\nadd_layers( model,\n Reshape((3, 32, 32), input_shape=(3*32*32,)),\n *MyConv2D(32, 3),\n MaxPool2D(),\n *MyConv2D(64, 3),\n MaxPool2D(),\n Reshape((-1,)),\n Dense(units=1024, activation=\"relu\"),\n Dropout(rate=0.4),\n Dense(units=10, activation=\"softmax\")\n)\nmodel.compile(loss='categorical_crossentropy',\n optimizer='adam',\n metrics=['accuracy'])\nmodel.fit(preprocessed_train_X, train_Y, validation_data=preprocessed_validation_data, \n batch_size=100, epochs=10)\nrtn = model.evaluate(*preprocessed_test_data)\nprint(\"\\ntest accuracy=\", rtn[1])", "Train on 50000 samples, validate on 1000 samples\nEpoch 1/10\n50000/50000 [==============================] - 19s - loss: 1.5286 - acc: 0.4543 - val_loss: 1.1498 - val_acc: 0.6020\nEpoch 2/10\n50000/50000 [==============================] - 19s - loss: 1.0738 - acc: 0.6200 - val_loss: 1.0352 - val_acc: 0.6340\nEpoch 3/10\n50000/50000 [==============================] - 19s - loss: 0.8797 - acc: 0.6888 - val_loss: 0.9085 - val_acc: 0.6720\nEpoch 4/10\n50000/50000 [==============================] - 18s - loss: 0.7187 - acc: 0.7481 - val_loss: 0.8152 - val_acc: 0.7040\nEpoch 5/10\n50000/50000 [==============================] - 19s - loss: 0.5719 - acc: 0.7980 - val_loss: 0.8395 - val_acc: 0.7110\nEpoch 6/10\n50000/50000 [==============================] - 19s - loss: 0.4424 - acc: 0.8460 - val_loss: 0.8739 - val_acc: 0.7140\nEpoch 7/10\n50000/50000 [==============================] - 18s - loss: 0.3275 - acc: 0.8860 - val_loss: 0.9208 - val_acc: 0.7170\nEpoch 8/10\n50000/50000 [==============================] - 18s - loss: 0.2517 - acc: 0.9131 - val_loss: 0.9607 - val_acc: 0.7130\nEpoch 9/10\n50000/50000 [==============================] - 19s - loss: 0.1930 - acc: 0.9334 - val_loss: 1.1707 - val_acc: 0.7090\nEpoch 10/10\n50000/50000 [==============================] - 18s - loss: 0.1592 - acc: 0.9462 - val_loss: 1.0804 - val_acc: 0.7220\n9000/9000 [==============================] - 1s \n\ntest accuracy= 0.701555555556\n" ], [ "import keras\nfrom keras.layers import Dense, Activation, Conv2D, MaxPool2D, Reshape, Dropout, BatchNormalization\n# 先處理資料\n\n#GCN\ntrain_X_mean = np.mean(train_X, axis=0, keepdims=True)\ntrain_X_std = np.std(train_X, axis=0, keepdims=True)\npreprocessed_train_X = (train_X-train_X_mean)/train_X_std\npreprocessed_test_X = (test_X-train_X_mean)/train_X_std\npreprocessed_validation_data = (preprocessed_test_X[:1000], test_Y[:1000])\npreprocessed_test_data = (preprocessed_test_X[1000:], test_Y[1000:])\n\n\ndef MyConv2D(filters, kernel_size, **kwargs):\n return (Conv2D(filters=filters, kernel_size=kernel_size, \n padding='same', data_format='channels_first', **kwargs),\n BatchNormalization(axis=1),\n Activation(\"relu\"))\ndef add_layers(model, *layers):\n for l in layers:\n model.add(l)\nmodel = Sequential()\nadd_layers( model,\n Reshape((3, 32, 32), input_shape=(3*32*32,)),\n *MyConv2D(32, 3),\n MaxPool2D(),\n *MyConv2D(64, 3),\n MaxPool2D(),\n Reshape((-1,)),\n Dense(units=1024, activation=\"relu\"), \n Dropout(0.4),\n Dense(units=10, activation=\"softmax\")\n)\nmodel.compile(loss='categorical_crossentropy',\n optimizer='adam',\n metrics=['accuracy'])\nmodel.fit(preprocessed_train_X, train_Y, validation_data=preprocessed_validation_data, \n batch_size=100, epochs=10)\nrtn = model.evaluate(*preprocessed_test_data)\nprint(\"\\ntest accuracy=\", rtn[1])", "Train on 50000 samples, validate on 1000 samples\nEpoch 1/10\n50000/50000 [==============================] - 57s - loss: 13.2463 - acc: 0.1721 - val_loss: 13.4396 - val_acc: 0.1640\nEpoch 2/10\n50000/50000 [==============================] - 56s - loss: 10.0067 - acc: 0.2496 - val_loss: 1.5723 - val_acc: 0.4020\nEpoch 3/10\n50000/50000 [==============================] - 57s - loss: 1.5455 - acc: 0.4333 - val_loss: 1.3153 - val_acc: 0.5190\nEpoch 4/10\n50000/50000 [==============================] - 57s - loss: 1.3747 - acc: 0.5027 - val_loss: 1.1882 - val_acc: 0.5760\nEpoch 5/10\n50000/50000 [==============================] - 56s - loss: 1.2656 - acc: 0.5405 - val_loss: 1.1008 - val_acc: 0.6060\nEpoch 6/10\n50000/50000 [==============================] - 57s - loss: 1.1868 - acc: 0.5729 - val_loss: 1.1156 - val_acc: 0.5970\nEpoch 7/10\n50000/50000 [==============================] - 56s - loss: 1.1118 - acc: 0.6008 - val_loss: 1.0348 - val_acc: 0.6390\nEpoch 8/10\n50000/50000 [==============================] - 57s - loss: 1.0523 - acc: 0.6248 - val_loss: 0.9923 - val_acc: 0.6450\nEpoch 9/10\n50000/50000 [==============================] - 57s - loss: 1.0043 - acc: 0.6419 - val_loss: 0.9261 - val_acc: 0.6870\nEpoch 10/10\n50000/50000 [==============================] - 56s - loss: 0.9496 - acc: 0.6605 - val_loss: 0.9329 - val_acc: 0.6740\n8960/9000 [============================>.] - ETA: 0s\ntest accuracy= 0.658888888889\n" ], [ "model.fit(preprocessed_train_X, train_Y, validation_data=preprocessed_validation_data, \n batch_size=100, epochs=10)\nrtn = model.evaluate(*preprocessed_test_data)\nprint(\"\\ntest accuracy=\", rtn[1])", "Train on 50000 samples, validate on 1000 samples\nEpoch 1/10\n50000/50000 [==============================] - 57s - loss: 0.9230 - acc: 0.6706 - val_loss: 0.8536 - val_acc: 0.6970\nEpoch 2/10\n50000/50000 [==============================] - 56s - loss: 0.8806 - acc: 0.6831 - val_loss: 0.8638 - val_acc: 0.6960\nEpoch 3/10\n50000/50000 [==============================] - 57s - loss: 0.8451 - acc: 0.6952 - val_loss: 0.8458 - val_acc: 0.7070\nEpoch 4/10\n50000/50000 [==============================] - 57s - loss: 0.8141 - acc: 0.7065 - val_loss: 0.9286 - val_acc: 0.6890\nEpoch 5/10\n50000/50000 [==============================] - 56s - loss: 0.7863 - acc: 0.7146 - val_loss: 0.8840 - val_acc: 0.6960\nEpoch 6/10\n50000/50000 [==============================] - 57s - loss: 0.7515 - acc: 0.7270 - val_loss: 0.8758 - val_acc: 0.7130\nEpoch 7/10\n50000/50000 [==============================] - 58s - loss: 0.7189 - acc: 0.7376 - val_loss: 0.8990 - val_acc: 0.6990\nEpoch 8/10\n50000/50000 [==============================] - 59s - loss: 0.6998 - acc: 0.7451 - val_loss: 0.8612 - val_acc: 0.7110\nEpoch 9/10\n50000/50000 [==============================] - 58s - loss: 0.6715 - acc: 0.7524 - val_loss: 0.8580 - val_acc: 0.7060\nEpoch 10/10\n50000/50000 [==============================] - 57s - loss: 0.6469 - acc: 0.7644 - val_loss: 0.8518 - val_acc: 0.7030\n8960/9000 [============================>.] - ETA: 0s\ntest accuracy= 0.706555555556\n" ], [ "def zca_whitening_matrix(X):\n \"\"\"\n Function to compute ZCA whitening matrix (aka Mahalanobis whitening).\n INPUT: X: [M x N] matrix.\n Rows: Variables\n Columns: Observations\n OUTPUT: ZCAMatrix: [M x M] matrix\n \"\"\"\n X = X.T\n # Covariance matrix [column-wise variables]: Sigma = (X-mu)' * (X-mu) / N\n sigma = np.cov(X, rowvar=True) # [M x M]\n # Singular Value Decomposition. X = U * np.diag(S) * V\n U,S,V = np.linalg.svd(sigma)\n # U: [M x M] eigenvectors of sigma.\n # S: [M x 1] eigenvalues of sigma.\n # V: [M x M] transpose of U\n # Whitening constant: prevents division by zero\n epsilon = 1e-5\n # ZCA Whitening matrix: U * Lambda * U'\n ZCAMatrix = np.dot(U, np.dot(np.diag(1.0/np.sqrt(S + epsilon)), U.T)) # [M x M]\n return ZCAMatrix.T\n# ZCAMatrix = zca_whitening_matrix(X0)\n# new_train_X= ((train_X-train_X_mean)/train_X_std) @ ZCAMatrix", "_____no_output_____" ], [ "# 參考 https://keras.io/preprocessing/image/\n\n# 輸入改成 tensor4\ntrain_X = train_X.reshape(-1, 3, 32, 32)\ntest_X = test_X.reshape(-1, 3, 32, 32)\n\ndef MyConv2D(filters, kernel_size, **kwargs):\n return (Conv2D(filters=filters, kernel_size=kernel_size, \n padding='same', data_format='channels_first', **kwargs),\n BatchNormalization(axis=1),\n Activation(\"relu\"))\nmodel = Sequential()\nadd_layers( model,\n *MyConv2D(32, 3, input_shape=(3,32,32)),\n MaxPool2D(),\n *MyConv2D(64, 3),\n MaxPool2D(),\n Reshape((-1,)),\n Dense(units=1024, activation=\"relu\"), \n Dropout(0.4),\n Dense(units=10, activation=\"softmax\")\n)\nmodel.compile(loss='categorical_crossentropy',\n optimizer='adam',\n metrics=['accuracy'])\n\n# 使用 keras 的功能\nfrom keras.preprocessing.image import ImageDataGenerator\ndatagen = ImageDataGenerator(\n featurewise_center=True,\n featurewise_std_normalization=True,\n zca_whitening=True,\n data_format=\"channels_first\")\n\n# compute quantities required for featurewise normalization\n# (std, mean, and principal components if ZCA whitening is applied)\ndatagen.fit(train_X)\n\n\np_train_X, p_train_Y = datagen.flow(train_X, train_Y, batch_size=len(train_X), shuffle=False).next()\n# 順序都沒變\nassert (p_train_Y == train_Y).all()\n\np_test_X, p_test_Y = datagen.flow(test_X, test_Y, batch_size=len(test_X), shuffle=False).next()\n# 順序都沒變\nassert (p_test_Y == test_Y).all()\n# 不需要這兩個\ndel p_train_Y, p_test_Y\n\n\np_validation_data = (p_test_X[:1000], test_Y[:1000])\np_test_data = (p_test_X[1000:], test_Y[1000:])\nmodel.fit(p_train_X, train_Y, validation_data=p_validation_data, \n batch_size=100, epochs=10)\nrtn = model.evaluate(*p_test_data)\nprint(\"\\ntest accuracy=\", rtn[1])\n", "Train on 50000 samples, validate on 1000 samples\nEpoch 1/10\n50000/50000 [==============================] - 59s - loss: 4.1387 - acc: 0.2103 - val_loss: 1.9535 - val_acc: 0.2730\nEpoch 2/10\n50000/50000 [==============================] - 58s - loss: 1.6590 - acc: 0.3671 - val_loss: 1.3423 - val_acc: 0.5450\nEpoch 3/10\n50000/50000 [==============================] - 59s - loss: 1.4440 - acc: 0.4645 - val_loss: 1.2077 - val_acc: 0.5660\nEpoch 4/10\n50000/50000 [==============================] - 59s - loss: 1.3028 - acc: 0.5202 - val_loss: 1.0343 - val_acc: 0.6300\nEpoch 5/10\n50000/50000 [==============================] - 58s - loss: 1.2161 - acc: 0.5540 - val_loss: 1.1199 - val_acc: 0.6130\nEpoch 6/10\n50000/50000 [==============================] - 58s - loss: 1.1516 - acc: 0.5752 - val_loss: 1.0667 - val_acc: 0.6310\nEpoch 7/10\n50000/50000 [==============================] - 57s - loss: 1.1062 - acc: 0.5923 - val_loss: 1.0645 - val_acc: 0.6160\nEpoch 8/10\n50000/50000 [==============================] - 57s - loss: 1.0729 - acc: 0.6030 - val_loss: 1.0230 - val_acc: 0.6330\nEpoch 9/10\n50000/50000 [==============================] - 58s - loss: 1.0288 - acc: 0.6178 - val_loss: 0.9881 - val_acc: 0.6540\nEpoch 10/10\n50000/50000 [==============================] - 57s - loss: 1.0024 - acc: 0.6277 - val_loss: 1.0081 - val_acc: 0.6570\n8960/9000 [============================>.] - ETA: 0s\ntest accuracy= 0.660666666667\n" ] ], [ [ "使用動態資料處理\n```python\n# fits the model on batches with real-time data augmentation:\ntrain_generator = datagen.flow(train_X, train_Y, batch_size=100, shuffle=False)\ntest_generator = datagen.flow(*test_data, batch_size=100, shuffle=False)\nmodel.fit_generator(train_generator,\n steps_per_epoch=len(train_X), \n #validation_data=datagen.flow(*validation_data, batch_size=100),\n #validation_steps=1000,\n epochs=10)\nrtn = model.evaluate_generator(test_generator, steps=9000)\n```", "_____no_output_____" ] ], [ [ "import keras\nfrom keras.layers import Dense, Activation, Conv2D, MaxPool2D, Reshape, Dropout, BatchNormalization\n# 輸入改成 tensor4\ntrain_X = train_X.reshape(-1, 3, 32, 32)\ntest_X = test_X.reshape(-1, 3, 32, 32)\n\ndef MyConv2D(filters, kernel_size, **kwargs):\n return (Conv2D(filters=filters, kernel_size=kernel_size, \n padding='same', data_format='channels_first', **kwargs),\n BatchNormalization(axis=1),\n Activation(\"elu\"))\nmodel = Sequential()\nadd_layers( model,\n *MyConv2D(64, 3, input_shape=(3,32,32)),\n *MyConv2D(64, 3),\n MaxPool2D(),\n *MyConv2D(128, 3),\n *MyConv2D(128, 3),\n MaxPool2D(),\n *MyConv2D(256, 3),\n *MyConv2D(256, 3),\n Reshape((-1,)),\n Dense(units=1024),\n BatchNormalization(),\n Activation(\"elu\"),\n Dropout(0.4),\n Dense(units=10, activation=\"softmax\")\n)\nmodel.compile(loss='categorical_crossentropy',\n optimizer='adam',\n metrics=['accuracy'])\n\n# 使用 keras 的功能\nfrom keras.preprocessing.image import ImageDataGenerator\ndatagen = ImageDataGenerator(\n featurewise_center=True,\n featurewise_std_normalization=True,\n zca_whitening=True,\n data_format=\"channels_first\")\n\n# compute quantities required for featurewise normalization\n# (std, mean, and principal components if ZCA whitening is applied)\ndatagen.fit(train_X)\n\n\np_train_X, p_train_Y = datagen.flow(train_X, train_Y, batch_size=len(train_X), shuffle=False).next()\n# 順序都沒變\nassert (p_train_Y == train_Y).all()\n\np_test_X, p_test_Y = datagen.flow(test_X, test_Y, batch_size=len(test_X), shuffle=False).next()\n# 順序都沒變\nassert (p_test_Y == test_Y).all()\n# 不需要這兩個\ndel p_train_Y, p_test_Y\n\n\np_validation_data = (p_test_X[:1000], test_Y[:1000])\np_test_data = (p_test_X[1000:], test_Y[1000:])\nmodel.fit(p_train_X, train_Y, validation_data=p_validation_data, \n batch_size=100, epochs=10)\nrtn = model.evaluate(*p_test_data)\nprint(\"\\ntest accuracy=\", rtn[1])\n", "Train on 50000 samples, validate on 1000 samples\nEpoch 1/10\n" ], [ "model.fit(p_train_X, train_Y, validation_data=p_validation_data, \n batch_size=100, epochs=10)\nrtn = model.evaluate(*p_test_data)", "_____no_output_____" ] ], [ [ "lasagne 中相同的\n```python\n_ = InputLayer(shape=(None, 3*32*32), input_var=input_var)\n_ = DropoutLayer(_, 0.2)\n_ = ReshapeLayer(_, ([0], 3, 32, 32))\n_ = conv(_, 96, 3)\n_ = conv(_, 96, 3)\n_ = MaxPool2DDNNLayer(_, 3, 2)\n_ = DropoutLayer(_, 0.5)\n_ = conv(_, 192, 3)\n_ = conv(_, 192, 3)\n_ = MaxPool2DDNNLayer(_, 3, 2)\n_ = DropoutLayer(_, 0.5)\n_ = conv(_, 192, 3)\n_ = conv(_, 192, 1)\n_ = conv(_, 10, 1)\n_ = Pool2DDNNLayer(_, 7, mode='average_exc_pad')\n_ = FlattenLayer(_)\nl_out = NonlinearityLayer(_, nonlinearity=lasagne.nonlinearities.softmax)\n```", "_____no_output_____" ] ], [ [ "import keras\nfrom keras.layers import Dense, Activation, Conv2D, MaxPool2D, Reshape, Dropout, BatchNormalization, GlobalAveragePooling2D\n# 輸入改成 tensor4\ntrain_X = train_X.reshape(-1, 3, 32, 32)\ntest_X = test_X.reshape(-1, 3, 32, 32)\n\ndef MyConv2D(filters, kernel_size, **kwargs):\n return (Conv2D(filters=filters, kernel_size=kernel_size, \n padding='same', data_format='channels_first', **kwargs),\n BatchNormalization(axis=1, momentum=0.9),\n Activation(\"relu\"))\nmodel = Sequential()\nadd_layers( model,\n Dropout(0.2, input_shape=(3,32,32)),\n *MyConv2D(96, 3),\n *MyConv2D(96, 3),\n MaxPool2D(3, 2),\n *MyConv2D(192, 3),\n *MyConv2D(192, 3),\n MaxPool2D(3, 2),\n Dropout(0.5),\n *MyConv2D(192, 3),\n *MyConv2D(192, 1),\n *MyConv2D(10, 1),\n GlobalAveragePooling2D(data_format='channels_first'),\n Activation(\"softmax\")\n)\nmodel.compile(loss='categorical_crossentropy',\n optimizer='adam',\n metrics=['accuracy'])\n\nmodel.fit(p_train_X, train_Y, validation_data=p_validation_data, \n batch_size=100, epochs=50)\nrtn = model.evaluate(*p_test_data)\nprint(\"\\ntest accuracy=\", rtn[1])\n", "_____no_output_____" ], [ "import keras\nfrom keras.layers import Dense, Activation, Conv2D, MaxPool2D, Reshape, Dropout, BatchNormalization, GlobalAveragePooling2D\n# 輸入改成 tensor4\ntrain_X = train_X.reshape(-1, 3, 32, 32)\ntest_X = test_X.reshape(-1, 3, 32, 32)\n\ndef MyConv2D(filters, kernel_size, **kwargs):\n return (Conv2D(filters=filters, kernel_size=kernel_size, \n padding='same', data_format='channels_first', **kwargs),\n BatchNormalization(axis=1, momentum=0.9),\n Activation(\"relu\"))\nmodel = Sequential()\nadd_layers( model,\n Dropout(0.2, input_shape=(3,32,32)),\n *MyConv2D(96, 3),\n *MyConv2D(96, 3),\n MaxPool2D(3, 2),\n *MyConv2D(192, 3),\n *MyConv2D(192, 3),\n MaxPool2D(3, 2),\n Dropout(0.5),\n *MyConv2D(192, 3),\n *MyConv2D(192, 1),\n *MyConv2D(10, 1),\n GlobalAveragePooling2D(data_format='channels_first'),\n Activation(\"softmax\")\n)\nmodel.compile(loss='categorical_crossentropy',\n optimizer='adam',\n metrics=['accuracy'])\n\n# 使用 keras 的功能\nfrom keras.preprocessing.image import ImageDataGenerator\ndatagen = ImageDataGenerator(\n featurewise_center=True,\n featurewise_std_normalization=True,\n zca_whitening=True,\n data_format=\"channels_first\")\n\n# compute quantities required for featurewise normalization\n# (std, mean, and principal components if ZCA whitening is applied)\ndatagen.fit(train_X)\n\n\np_train_X, p_train_Y = datagen.flow(train_X, train_Y, batch_size=len(train_X), shuffle=False).next()\n# 順序都沒變\nassert (p_train_Y == train_Y).all()\n\np_test_X, p_test_Y = datagen.flow(test_X, test_Y, batch_size=len(test_X), shuffle=False).next()\n# 順序都沒變\nassert (p_test_Y == test_Y).all()\n# 不需要這兩個\ndel p_train_Y, p_test_Y\n\n\np_validation_data = (p_test_X[:1000], test_Y[:1000])\np_test_data = (p_test_X[1000:], test_Y[1000:])\nmodel.fit(p_train_X, train_Y, validation_data=p_validation_data, \n batch_size=100, epochs=10)\nrtn = model.evaluate(*p_test_data)\nprint(\"\\ntest accuracy=\", rtn[1])\n", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import sys\nsys.path.insert(1,\"/home1/07064/tg863631/anaconda3/envs/CbrainCustomLayer/lib/python3.6/site-packages\") #work around for h5py\nfrom cbrain.imports import *\nfrom cbrain.cam_constants import *\nfrom cbrain.utils import *\nfrom cbrain.layers import *\nfrom cbrain.data_generator import DataGenerator\nimport tensorflow as tf\nfrom tensorflow import math as tfm\nfrom tensorflow.keras.layers import *\nfrom tensorflow.keras.models import *\n# import tensorflow_probability as tfp\nimport xarray as xr\nimport numpy as np\nfrom cbrain.model_diagnostics import ModelDiagnostics\nimport matplotlib as mpl\nimport matplotlib.pyplot as plt\nimport matplotlib.image as imag\nimport scipy.integrate as sin\nimport matplotlib.ticker as mticker\nimport pickle\nfrom tensorflow.keras import layers\nfrom tensorflow.keras.losses import *\nfrom tensorflow.keras.callbacks import ModelCheckpoint, EarlyStopping\nimport datetime\nfrom cbrain.climate_invariant import *\nimport yaml\nfrom cbrain.imports import *\nfrom cbrain.utils import *\nfrom cbrain.normalization import *\nimport h5py\nfrom sklearn.preprocessing import OneHotEncoder", "_____no_output_____" ] ], [ [ "## Preprocess the data and dump to a new file", "_____no_output_____" ] ], [ [ "DATA_PATH = '/fast/ankitesh/data/'\nTRAINFILE = 'CI_SP_M4K_train_shuffle.nc'\nVALIDFILE = 'CI_SP_M4K_valid.nc'\nNORMFILE = 'CI_SP_M4K_NORM_norm.nc'\npercentile_path='/export/nfs0home/ankitesg/data/percentile_data.pkl'\ndata_name='M4K'\nbin_size = 1000\nscale_dict = load_pickle('/export/nfs0home/ankitesg/CBrain_project/CBRAIN-CAM/nn_config/scale_dicts/009_Wm2_scaling.pkl')", "_____no_output_____" ], [ "percentile_bins = load_pickle(percentile_path)['Percentile'][data_name]\nenc = OneHotEncoder(sparse=False)\nclasses = np.arange(bin_size+2)\nenc.fit(classes.reshape(-1,1))", "_____no_output_____" ], [ "data_ds = xr.open_dataset(f\"{DATA_PATH}{TRAINFILE}\")\nn = data_ds['vars'].shape[0]", "_____no_output_____" ], [ "data_ds", "_____no_output_____" ], [ "coords = list(data_ds['vars'].var_names.values)\ncoords = coords + ['PHQ_BIN']*30+['TPHYSTND_BIN']*30+['FSNT_BIN','FSNS_BIN','FLNT_BIN','FLNS_BIN']", "_____no_output_____" ], [ "def _transform_to_one_hot(Y):\n '''\n return shape = batch_size X 64 X bin_size\n '''\n\n Y_trans = []\n out_vars = ['PHQ','TPHYSTND','FSNT', 'FSNS', 'FLNT', 'FLNS']\n var_dict = {}\n var_dict['PHQ'] = Y[:,:30]\n var_dict['TPHYSTND'] = Y[:,30:60]\n var_dict['FSNT'] = Y[:,60]\n var_dict['FSNS'] = Y[:,61]\n var_dict['FLNT'] = Y[:,62]\n var_dict['FLNS'] = Y[:,63]\n perc = percentile_bins\n for var in out_vars[:2]:\n all_levels_one_hot = []\n for ilev in range(30):\n bin_index = np.digitize(var_dict[var][:,ilev],perc[var][ilev])\n one_hot = enc.transform(bin_index.reshape(-1,1))\n all_levels_one_hot.append(one_hot)\n var_one_hot = np.stack(all_levels_one_hot,axis=1) \n Y_trans.append(var_one_hot)\n for var in out_vars[2:]:\n bin_index = np.digitize(var_dict[var][:], perc[var])\n one_hot = enc.transform(bin_index.reshape(-1,1))[:,np.newaxis,:]\n Y_trans.append(one_hot)\n\n Y_concatenated = np.concatenate(Y_trans,axis=1)\n return Y_concatenated", "_____no_output_____" ], [ "inp_vars = ['QBP','TBP','PS', 'SOLIN', 'SHFLX', 'LHFLX']\ninp_coords = coords[:64]\nout_coords = coords[64:128]\nbin_coords = list(range(bin_size+2))", "_____no_output_____" ], [ "all_data_arrays = []\nbatch_size = 4096\nnorm_ds = xr.open_dataset(f'{DATA_PATH}{NORMFILE}')\noutput_transform = DictNormalizer(norm_ds, ['PHQ','TPHYSTND','FSNT', 'FSNS', 'FLNT', 'FLNS'], scale_dict)\nfor i in range(0,n,batch_size):\n all_vars = data_ds['vars'][i:i+batch_size]\n inp_vals = all_vars[:,:64]\n out_vals = all_vars[:,64:128]\n out_vals = output_transform.transform(out_vals)\n one_hot = _transform_to_one_hot(out_vals)\n sample_coords = list(range(i,i+all_vars.shape[0]))\n x3 = xr.Dataset(\n {\n \"X\": ((\"sample\", \"inp_coords\"),inp_vals),\n \"Y_raw\":((\"sample\",\"out_cords\"),out_vals),\n \"Y\": ((\"sample\", \"out_coords\",\"bin_index\"), one_hot),\n },\n coords={\"sample\": sample_coords, \"inp_coords\": inp_coords,\"out_coords\":out_coords,\"bin_index\":bin_coords},\n )\n all_data_arrays.append(x3)\n if(int(i/batch_size+1)%100 == 0):\n print(\"saving this batch\")\n final_da = xr.combine_by_coords(all_data_arrays)\n final_da.to_netcdf(f'/scratch/ankitesh/data/new_data_for_v2_{int(i/batch_size+1)}.nc')\n all_data_arrays = []\n print(int(i/batch_size), end='\\r')", "saving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\n10343\r" ], [ "final_da = xr.combine_by_coords(all_data_arrays)", "_____no_output_____" ], [ "final_da", "_____no_output_____" ], [ "data_ds = xr.open_dataset(f\"{DATA_PATH}{VALIDFILE}\")\nn = data_ds['vars'].shape[0]", "_____no_output_____" ], [ "coords = list(data_ds['vars'].var_names.values)\ncoords = coords + ['PHQ_BIN']*30+['TPHYSTND_BIN']*30+['FSNT_BIN','FSNS_BIN','FLNT_BIN','FLNS_BIN']", "_____no_output_____" ], [ "all_data_arrays = []\nbatch_size = 4096\nnorm_ds = xr.open_dataset(f'{DATA_PATH}{NORMFILE}')\noutput_transform = DictNormalizer(norm_ds, ['PHQ','TPHYSTND','FSNT', 'FSNS', 'FLNT', 'FLNS'], scale_dict)\nfor i in range(0,n,batch_size):\n all_vars = data_ds['vars'][i:i+batch_size]\n inp_vals = all_vars[:,:64]\n out_vals = all_vars[:,64:128]\n out_vals = output_transform.transform(out_vals)\n one_hot = _transform_to_one_hot(out_vals)\n sample_coords = list(range(i,i+all_vars.shape[0]))\n x3 = xr.Dataset(\n {\n \"X\": ((\"sample\", \"inp_coords\"),inp_vals),\n \"Y_raw\":((\"sample\",\"out_cords\"),out_vals),\n \"Y\": ((\"sample\", \"out_coords\",\"bin_index\"), one_hot),\n },\n coords={\"sample\": sample_coords, \"inp_coords\": inp_coords,\"out_coords\":out_coords,\"bin_index\":bin_coords},\n )\n all_data_arrays.append(x3)\n if(int(i/batch_size+1)%100 == 0):\n print(\"saving this batch\")\n final_da = xr.combine_by_coords(all_data_arrays)\n final_da.to_netcdf(f'/scratch/ankitesh/data/new_data_valid_for_v2_{int(i/batch_size+1)}.nc')\n all_data_arrays = []\n print(int(i/batch_size), end='\\r')\nfinal_da = xr.combine_by_coords(all_data_arrays)\nfinal_da.to_netcdf(f'/scratch/ankitesh/data/new_data_valid_for_v2_{int(i/batch_size+1)}.nc')", "saving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\nsaving this batch\n" ] ] ]

[ "code", "markdown", "code" ]

[ [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "CC-BY-4.0" ]

[ "CC-BY-4.0" ]

[ "CC-BY-4.0" ]

[ [ [ "empty" ] ] ]

[ "empty" ]

[ [ "empty" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import pandas as pd", "_____no_output_____" ], [ "df = pd.read_csv('Datasets/direct_marketing.csv')\ndf.head()", "_____no_output_____" ] ], [ [ "## Slicing", "_____no_output_____" ] ], [ [ "df.recency\ndf['recency']\ndf[['recency']]\ndf.loc[:,'recency']\ndf.loc[:,['recency']]\ndf.iloc[:,0]\ndf.iloc[:,[0]]", "_____no_output_____" ] ], [ [ "## Boolean Indexing", "_____no_output_____" ] ], [ [ "df.recency < 7\ndf[ df.recency < 7 ]\ndf[ (df.recency < 7) & (df.newbie == 0) ]\n", "_____no_output_____" ], [ "df[df.recency < 7] = -100", "_____no_output_____" ], [ "ordered_satisfaction = ['Very Unhappy', 'Unhappy', 'Neutral', 'Happy', 'Very Happy']\ndf = pd.DataFrame({'satisfaction':['Mad', 'Happy', 'Unhappy', 'Neutral']})\ndf.satisfaction = df.satisfaction.astype(\"category\",\n ordered=True,\n categories=ordered_satisfaction\n).cat.codes", "/home/sasha/miniconda3/envs/pwpyfds/lib/python3.7/site-packages/ipykernel_launcher.py:5: FutureWarning: specifying 'categories' or 'ordered' in .astype() is deprecated; pass a CategoricalDtype instead\n \"\"\"\n" ], [ "df['satisfaction']", "_____no_output_____" ], [ "df = pd.DataFrame({'vertebrates':[\n... 'Bird',\n... 'Bird',\n... 'Mammal',\n... 'Fish',\n... 'Amphibian',\n... 'Reptile',\n... 'Mammal',\n... ]})", "_____no_output_____" ], [ "df.vertebrates.astype('category').cat.codes, df.vertebrates.astype('category').cat.categories", "_____no_output_____" ], [ "pd.get_dummies(df, columns=['vertebrates'])", "_____no_output_____" ] ], [ [ "## Feature Representation", "_____no_output_____" ], [ "### Pure Textual Features", "_____no_output_____" ] ], [ [ "from sklearn.feature_extraction.text import CountVectorizer\n\t\ncorpus = [\n \"Authman ran faster than Harry because he is an athlete.\",\n \"Authman and Harry ran faster and faster.\",\n ]\n\t", "_____no_output_____" ], [ "bow = CountVectorizer()\nX = bow.fit_transform(corpus) # Sparse Matrix", "_____no_output_____" ], [ "bow.get_feature_names()", "_____no_output_____" ], [ "X.toarray()", "_____no_output_____" ] ], [ [ "# Graphical Features", "_____no_output_____" ] ], [ [ "%matplotlib inline\nimport matplotlib.pyplot as plt\nfrom imageio import imread\n\t\n# Load the image up\nimg = imread('imageio:chelsea.png')\n\t\n# Is the image too big? Resample it down by an order of magnitude\nimg = img[::2, ::2]\n\t\n# Scale colors from (0-255) to (0-1), then reshape to 1D array per pixel, e.g. grayscale\n# If you had color images and wanted to preserve all color channels, use .reshape(-1,3)\nX = (img / 255.0).reshape(-1)", "_____no_output_____" ], [ "plt.imshow(img)", "_____no_output_____" ], [ "X.shape", "_____no_output_____" ] ], [ [ "## Audio Features", "_____no_output_____" ] ], [ [ "import scipy.io.wavfile as wavfile\n\t\nsample_rate, audio_data = wavfile.read('sound.wav')\nprint(audio_data)", "_____no_output_____" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Docutils", "_____no_output_____" ], [ "## Presentation", "_____no_output_____" ], [ "Click [__here__] (youtube link) for the video presentation", "_____no_output_____" ], [ "## Summary of Support Files", "_____no_output_____" ], [ "- `demo.ipynb`: the notebook containing this tutorial code\n- `test.csv`: a small file data used in the tutorial code", "_____no_output_____" ], [ "## Installation Instructions", "_____no_output_____" ], [ "Use `!pip install docutils` to install the `docutils` package. Next, use `import docutils`to import the package into your notebook.\n\nFor example, to import specific modules from `docutils` package use the following line of code:\n `from docutils import core, io`\n \nBelow is a list of modules and subpackages as apart of the `docutils` package:", "_____no_output_____" ], [ "## Guide", "_____no_output_____" ], [ "__docutils 0.17.1 version__\n\n- Author: David Goodger\n- Contact: goodger@python.org\n\n[Docutils](https://pypi.org/project/docutils/) is an open-source, modular text processing system for processing plaintext documentation into a more useful format. Formats include HTML, man-pages, OpenDocument, LaTeX, or XML.\n\nDocutils supports reStructuredText for input, an easy-to-read, what-you-see-is-what-you-get plaintext markup syntax.\n\nDocutils is short for \"Python Documentation Utilities\".", "_____no_output_____" ], [ "Support for the following sources has been implemented:\n - Standalone files\n - `PEPs (Python Enhancement Proposals)`\n \nSupport for these sources is currently being developed:\n - Inline documentation\n - Wikis\n - Email and more", "_____no_output_____" ], [ "Docutils Distribution Consists of:\n - the `docutils` package (or library)\n - front-end tools\n - test suite\n - documentation.\n ", "_____no_output_____" ], [ "## Notable docutils Modules & Subpackages\n-----------------------------\nModule | Definition\n------------- | -------------\n__core__ | Contains the ``Publisher`` class and ``publish_()``convenience functions\n__io__ | Provides a uniform API for low-level input and output\n__nodes__ | Docutils document tree (doctree) node class library\n\n\n-----------------------------\nSubpackages | Definition\n------------- | -------------\n**languages** | Language-specific mappings of terms\n**parsers** | Syntax-specific input parser modules or packages\n**readers** | Context-specific input handlers which understand the data source and manage a parser\n", "_____no_output_____" ], [ "Below is an overview of the `docutils` package:\n![alt text](docutils.png \"Docutils\")\n", "_____no_output_____" ], [ "## Main Use Applications of Package", "_____no_output_____" ], [ "The reStructured Text component of the `docutils` package makes it easy to convert between different formats, especially from plain text to a static website. It is unique because it is extensible. Better than simpler markups. \n\nAdditionally, users can pair `docutils` with `Sphinx` to convert text to html. The `Sphinx` package is built on the `docutils` package. The docutils parser creates the parse tree as a representation of the text in the memory for the Sphinx application and .rst environment.", "_____no_output_____" ] ] ]

[ "markdown" ]

[ [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import numpy # noqa: F401", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import numpy as np\nimport pandas as pd\nfrom matplotlib import pyplot as plt\nfrom scipy.spatial.transform import Rotation as R\nimport copy\n\n# Let's write an expanded-ensemble Sampler() class\n\nclass EESampler_RigidThreeParticle(object):\n \"\"\"An expanded-ensemble Sampler class for a rigid-triangle 3-particle/3-restraint system.\"\"\"\n \n def __init__(self, k_values=[0.0, 0.5, 1.0, 2.0, 5.0, 10., 15., 20., 50., 100., 200., 400., 800.], L=8.0,\n x0=np.array([[0.,0.,0.],[1.,0.,0.], [2.,1.,0.]]),\n a1=np.array([[0.,0.,0.],[1.,0.,0.], [2.,1.,0.]])):\n \"\"\"Initialize the class.\n \n INPUT\n k -- a np.array() of force constant values (kJ/nm^2)\n L -- the length of the cubic box (nm), extending from -L/2 to L/2.\n x0 -- the initial positions of the particles (np.array of shape (3,3))\n a1 -- position of the harmonic anchors (np.array of shape (3,3))\n \"\"\"\n \n self.k_values = np.array(k_values) # kJ/mol/nm^2)\n self.L = L # nm\n self.x0 = x0 # np.array with shape (3,3) and units nm\n self.a1 = a1 # \"\n \n # Calculate distance between particles 1 and 2\n self.d0 = self.distance(x0[0,:], x0[1,:]) # the initial distance between p2 - p1, as a reference\n self.d = copy.copy(self.d0)\n print('self.d', self.d)\n \n # Calculate the altitude (height), c, of triangle where p1-p2 is the base\n ### 1. First calculate the area of the triangle as A = (1/2)||v \\cross w||\n p1, p2, p3 = x0[0,:], x0[1,:], x0[2,:] \n v, w = p2-p1, p3-p1\n area = 0.5*np.linalg.norm( np.cross(v,w) )\n ### 2. Then since A = 1/2 * base * height, the height is c = 2*area/base\n self.c0 = 2.0 * area / np.linalg.norm(v)\n self.c = copy.copy(self.c0)\n print('self.c', self.c)\n \n # calculate e, the projection of p3-p1 = w in the p2-p1 = v direction\n unit_vec_along_p12 = v / self.d\n self.e = np.abs(np.dot(w,unit_vec_along_p12))\n print('self.e', self.e)\n \n self.d3 = np.linalg.norm(w) # the initial distance between p3 - p1, as a reference\n \n \n self.k_index = 0 # thermodynamic index\n\n self.n_ensembles = self.k_values.shape[0]\n\n self.x = copy.copy(self.x0)\n self.U = self.energy(self.x, self.k_values[self.k_index])\n \n \n ### Monte Carlo and Wang-Landau (WL) settings\n self.dx = 0.2 # Gaussian translation step size \n self.dtheta = 0.2 # Gaussian step size for angular (in radians)\n \n self.RT = 2.479 # in kJ/mol at 298 K\n \n self.all_pbc_shifts = []\n for i in [-L, 0, L]:\n for j in [-L, 0, L]:\n for k in [-L, 0, L]:\n self.all_pbc_shifts.append([i,j,k])\n self.all_pbc_shifts = np.array(self.all_pbc_shifts)\n # print('self.all_pbc_shifts', self.all_pbc_shifts)\n \n self.wl_increment = 5.0 # in kT\n self.wl_scaling = 0.5 # (0.5 is a la R. E. Belardinelli, and V. D. Pereyra, JCP 2007), 0.8 is MRS\n self.flatness = 0.8 # if all histogram values are within flatness*100\n # percent of the mean counts, decrease wl_increment and reset histogram\n self.wl_increment_freq = 10 # frequency to update wl sampling\n self.g = np.zeros(self.n_ensembles) # bias energies\n self.h = np.zeros(self.n_ensembles) # histogram counts\n \n self.use_1_over_t = False # switch to updating increment as c_1_over_t/(nsteps) when\n self.c_1_over_t = 10.0 # the wl_increment < c_1_over_t/(nsteps) \n # (a la R. E. Belardinelli, and V. D. Pereyra, JCP 2007)\n self.has_switched_to_1_over_t = False\n self.steps_before_1_over_t = 10000 # wait at least these number of steps before switching to 1/t\n self.print_every = 10000 \n self.traj_every = 100 \n \n def theory_dg_in_kT(self, verbose=True):\n \"\"\"Returns the theoretical value for the 3-restraint potential at all k_values.\"\"\"\n if verbose:\n print('self.d', self.d)\n print('self.c', self.c)\n print('self.e', self.e)\n print('self.L', self.L)\n\n kc_coeff = 1.0 + (self.e/self.d)**2.0\n if verbose:\n print('kc_coeff', kc_coeff)\n kp1_coeff = 1.0 + (self.c**2 + self.e**2)/(self.d**2.0)\n if verbose:\n print('kp1_coeff', kp1_coeff)\n\n '''\n theory_dG_in_kT = -1.0*( 3.0/2.0*np.log(2.0*np.pi*ee.RT/(3.0*ee.k_values[1:])) \\ # translational\n + 1.0/2.0*np.log(2.0*np.pi*ee.RT/(2.0*ee.k_values[1:])) \\ # rot about d\n + 1.0/2.0*np.log(2.0*np.pi*ee.RT/(kc_coeff*ee.k_values[1:])) \\ # rot about c\n + 1.0/2.0*np.log(2.0*np.pi*ee.RT/(kp1_coeff*ee.k_values[1:])) \\ # rot out of page\n - np.log( ee.L**3 * 8.0 * (np.pi**2) * (ee.d)**2 * ee.c ) )\n '''\n\n theory_dG_in_kT = -1.0*( 3.0/2.0*np.log(2.0*np.pi*self.RT/(3.0*self.k_values[1:])) \\\n + 1.0/2.0*np.log(2.0*np.pi*self.RT/(self.k_values[1:])) \\\n + 1.0/2.0*np.log(2.0*np.pi*self.RT/(kc_coeff*self.k_values[1:])) \\\n + 1.0/2.0*np.log(2.0*np.pi*self.RT/(kp1_coeff*self.k_values[1:])) \\\n - np.log( self.L**3 * 8.0 * (np.pi**2) * (self.d)**2 * self.c ) )\n return theory_dG_in_kT\n \n def energy(self, x, k):\n \"\"\"Returns the energy of the harmonic potential in units kJ/mol, for position x.\"\"\"\n return 0.5*k* np.sum((x-self.a1)*(x-self.a1))\n \n def distance(self, p1, p2):\n \"\"\"Return the distance between two particles (np.array of shape (3,))\"\"\"\n return np.sqrt( np.dot(p2-p1,p2-p1) ) \n \n def sample(self, nsteps, verbose=True):\n \"\"\"Perform WL expanded ensemble sampling with MC position moves.\"\"\"\n \n # store transition counts\n T_counts = np.zeros( (self.n_ensembles, self.n_ensembles) )\n accepted = 0\n data = []\n \n for step in range(nsteps):\n\n ########################\n ### MC moves\n \n # Before we do any moves, we need to reassemble the rod if it has\n # disjointed from any periodic box separations.\n p1, p2, p3 = self.x[0,:], self.x[1,:], self.x[2,:]\n \n tol = 1e-3\n # print('p1, p2', p1, p2, 'self.distance(p1, p2)', self.distance(p1, p2), 'self.d', self.d)\n p2_distance_deviation = np.abs(self.distance(p1, p2) - self.d0)\n p3_distance_deviation = np.abs(self.distance(p1, p3) - self.d3)\n if p2_distance_deviation + p3_distance_deviation > tol:\n # print(step, 'points out of the box. Reassembling...') \n ## Find all images of the second and third particles\n all_images_p2 = np.tile(p2, 27).reshape(27, 3) + self.all_pbc_shifts\n all_images_p3 = np.tile(p3, 27).reshape(27, 3) + self.all_pbc_shifts\n ### select the image that is the closest to particle 1 (note: this only works of self.d < L/2)\n p2_displacements = all_images_p2 - np.tile(p1, 27).reshape(27, 3)\n p3_displacements = all_images_p3 - np.tile(p1, 27).reshape(27, 3)\n # print('displacements', displacements)\n p2_sq_distances = np.sum(p2_displacements * p2_displacements, axis=1)\n p3_sq_distances = np.sum(p3_displacements * p3_displacements, axis=1)\n # print('sq_distances', sq_distances)\n p2_closest_index = np.argmin(p2_sq_distances) \n p2 = all_images_p2[p2_closest_index,:]\n p3_closest_index = np.argmin(p3_sq_distances) \n p3 = all_images_p3[p3_closest_index,:]\n x_assembled = np.array([list(p1),list(p2),list(p3)])\n # print('step', step, 'x_assembled', x_assembled)\n \n # For debugging, store the distance betwen the poarticles for the reassembled rod\n self.d = self.distance(x_assembled[0,:], x_assembled[1,:])\n \n # For debugging, store the triangle altitude c\n v, w = p2-p1, p3-p1\n self.c = np.linalg.norm( np.cross(v,w) ) / np.linalg.norm(v)\n \n # propose a MC translation move\n nudge = np.random.randn(3)\n x_new = x_assembled + self.dx*np.tile(nudge, 3).reshape(3,3)\n \n # propose a rotation move\n \n ## get a random vector on the unit sphere\n vec = 2.0*np.random.random(3) - 1.0\n while np.dot(vec,vec) > 1.0:\n vec = 2.0*np.random.random(3) - 1.0\n vec = vec/np.sqrt( np.dot(vec,vec) ) \n \n ## Find the rotation transformation about this axis\n rotation_radians = self.dtheta*np.random.randn()\n rotation_vector = rotation_radians * vec\n rotation = R.from_rotvec(rotation_vector)\n \n ## Translate the rigid rod to the origin\n midpoint = (x_new[0,:] + x_new[1,:] + x_new[2,:])/3.0\n x_new -= np.tile(midpoint, 3).reshape(3,3)\n ## Rotate it\n x_new_rotated = rotation.apply(x_new)\n # print(x_new_rotated)\n ## Translate it back\n x_new_rotated += np.tile(midpoint, 3).reshape(3,3)\n \n # if any coordinate of x_new goes outside the box, wrap it back in:\n x_new = ((x_new_rotated + np.array([self.L/2., self.L/2., self.L/2.])) % self.L) - np.array([self.L/2., self.L/2., self.L/2.]) \n U_new = self.energy(x_new, self.k_values[self.k_index])\n # print('step', step, 'self.x', 'x_new', self.x, x_new)\n # print('\\tself.U', 'U_new', self.U, U_new)\n \n \n # accept MC move according to the metropolis criterion\n accept = False\n P_accept = min(1., np.exp(-(U_new - self.U)/self.RT))\n if np.random.rand() < P_accept:\n accept = True\n if accept:\n accepted += 1\n self.x = x_new\n self.U = U_new\n\n ########################\n ### WL moves\n ### next, update the WL histogams and propose a WL move\n \n if (step%self.wl_increment_freq == 1):\n\n # WL histograms\n self.g[self.k_index] -= self.wl_increment\n self.h[self.k_index] += 1.0\n\n # reset the bias to i=0 reference\n self.g -= self.g[0]\n\n # attempt a move to a neighboring ensemble\n wl_accept = False\n if np.random.rand() < 0.5:\n k_index_new = self.k_index + 1\n else:\n k_index_new = self.k_index - 1\n\n if (k_index_new >= 0) and (k_index_new < self.n_ensembles):\n energies = self.energy(self.x, self.k_values) \n P_wl_accept = min(1., np.exp( -(energies[k_index_new]/self.RT - self.g[k_index_new] - energies[self.k_index]/self.RT + self.g[self.k_index] )))\n if np.random.rand() < P_wl_accept: #print(f'2 - is now {particles[x]}')\n wl_accept = True \n if wl_accept:\n T_counts[self.k_index, k_index_new] += 1.0\n self.k_index = k_index_new\n else:\n T_counts[self.k_index,self.k_index] += 1.0\n\n\n # check if the histogram is flat enough\n mean_counts = self.h.mean()\n\n which_are_flat_enough = (self.h > self.flatness*mean_counts)*(self.h < (2.0-self.flatness)*mean_counts)\n if step%self.print_every == 0:\n print('h', self.h, 'mean_counts', mean_counts)\n print('which_are_flat_enough', which_are_flat_enough)\n\n if step >= self.steps_before_1_over_t:\n if (self.has_switched_to_1_over_t == False) and (self.wl_increment < self.c_1_over_t/float(step)):\n print('#### Switching to 1/t method ( wl_increment =', self.wl_increment, ') ####')\n self.has_switched_to_1_over_t = True\n\n if np.sum( which_are_flat_enough.astype(int) ) == self.n_ensembles:\n if self.has_switched_to_1_over_t:\n self.wl_increment = self.c_1_over_t/float(step)\n else:\n self.wl_increment *= self.wl_scaling\n self.h = np.zeros(self.n_ensembles)\n\n # print a status report\n if step%self.print_every == 0:\n print('step', step, '| λ index', self.k_index, '| wl_increment =', self.wl_increment, 'kT')\n if not verbose: # just post first and last ensemble info\n print('%8d\\t%8d\\t%3.4f'%(0, self.h[0], self.g[0]))\n print('%8d\\t%8d\\t%3.4f'%(self.n_ensembles, self.h[-1], self.g[-1]))\n if verbose:\n print('# ensemble\\tk_value\\thistogram\\tg (kT)')\n for k in range(self.n_ensembles):\n outstr = '%8d\\t%4.1f\\t%8d\\t%3.4f'%(k, self.k_values[k], self.h[k], self.g[k])\n if k == self.k_index:\n outstr += ' <<'\n print(outstr)\n print()\n\n # store sample in trajectory\n if step%self.traj_every == 0:\n dhdl = self.energy(self.x, self.k_values) - self.energy(self.x, self.k_index)\n data.append([step, self.k_index,\n self.x[0,0], self.x[0,1], self.x[0,2],\n self.x[1,0], self.x[1,1], self.x[1,2],\n self.x[2,0], self.x[2,1], self.x[2,2],\n self.d, self.c,\n self.wl_increment, self.g[-1], dhdl, T_counts])\n\n\n trajectory = pd.DataFrame(data, columns=['step', 'lambda',\n 'x1', 'y1', 'z1', 'x2', 'y2', 'z2', 'x3', 'y3', 'z3', 'distance', 'height',\n 'increment','free_energy','dhdl', 'transition_matrix'])\n print(accepted)\n return trajectory\n\n\n ", "_____no_output_____" ], [ "ee = EESampler_RigidThreeParticle()\n", "self.d 1.0\nself.c 1.0\nself.e 2.0\n" ], [ "ee = EESampler_RigidThreeParticle()\n\nnsteps = 2000000\ntraj = ee.sample(nsteps)\n# print(traj['dhdl'])\n\nstep = traj.loc[:,'step'].values\np1 = np.zeros((step.shape[0], 3))\np1[:,0] = traj.loc[:,'x1'].values\np1[:,1] = traj.loc[:,'y1'].values\np1[:,2] = traj.loc[:,'z1'].values\np2 = np.zeros((step.shape[0], 3))\np2[:,0] = traj.loc[:,'x2'].values\np2[:,1] = traj.loc[:,'y2'].values\np2[:,2] = traj.loc[:,'z2'].values\nd = traj.loc[:,'distance'].values\nc = traj.loc[:,'height'].values\n\nimport matplotlib\nfrom matplotlib import pyplot as plt\nplt.figure(figsize=(10,4))\nplt.subplot(1,3,1)\nplt.plot(step, p1)\nplt.plot(step, p2)\nplt.subplot(1,3,2)\nplt.plot(step, d)\nplt.subplot(1,3,3)\nplt.plot(step, c)\n", "self.d 1.0\nself.c 1.0\nself.e 2.0\nstep 0 | λ index 0 | wl_increment = 5.0 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 0\t0.0000 <<\n 1\t 0.5\t 0\t0.0000\n 2\t 1.0\t 0\t0.0000\n 3\t 2.0\t 0\t0.0000\n 4\t 5.0\t 0\t0.0000\n 5\t10.0\t 0\t0.0000\n 6\t15.0\t 0\t0.0000\n 7\t20.0\t 0\t0.0000\n 8\t50.0\t 0\t0.0000\n 9\t100.0\t 0\t0.0000\n 10\t200.0\t 0\t0.0000\n 11\t400.0\t 0\t0.0000\n 12\t800.0\t 0\t0.0000\n\nstep 10000 | λ index 11 | wl_increment = 2.5 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 0\t0.0000\n 1\t 0.5\t 0\t-5.0000\n 2\t 1.0\t 0\t-10.0000\n 3\t 2.0\t 0\t-10.0000\n 4\t 5.0\t 0\t-10.0000\n 5\t10.0\t 6\t0.0000\n 6\t15.0\t 6\t0.0000\n 7\t20.0\t 5\t2.5000\n 8\t50.0\t 19\t2.5000\n 9\t100.0\t 25\t7.5000\n 10\t200.0\t 30\t10.0000\n 11\t400.0\t 33\t7.5000 <<\n 12\t800.0\t 33\t12.5000\n\nstep 20000 | λ index 10 | wl_increment = 2.5 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 113\t0.0000\n 1\t 0.5\t 106\t12.5000\n 2\t 1.0\t 103\t15.0000\n 3\t 2.0\t 102\t17.5000\n 4\t 5.0\t 100\t22.5000\n 5\t10.0\t 101\t45.0000\n 6\t15.0\t 101\t45.0000\n 7\t20.0\t 94\t62.5000\n 8\t50.0\t 96\t92.5000\n 9\t100.0\t 85\t140.0000\n 10\t200.0\t 68\t197.5000 <<\n 11\t400.0\t 51\t245.0000\n 12\t800.0\t 37\t285.0000\n\nstep 30000 | λ index 6 | wl_increment = 1.25 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 93\t0.0000\n 1\t 0.5\t 99\t5.0000\n 2\t 1.0\t 97\t10.0000\n 3\t 2.0\t 93\t17.5000\n 4\t 5.0\t 81\t37.5000\n 5\t10.0\t 71\t72.5000\n 6\t15.0\t 42\t108.7500 <<\n 7\t20.0\t 23\t142.5000\n 8\t50.0\t 20\t148.7500\n 9\t100.0\t 24\t156.2500\n 10\t200.0\t 46\t163.7500\n 11\t400.0\t 61\t182.5000\n 12\t800.0\t 82\t186.2500\n\nstep 40000 | λ index 6 | wl_increment = 0.625 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 0\t0.0000\n 1\t 0.5\t 0\t2.5000\n 2\t 1.0\t 0\t6.2500\n 3\t 2.0\t 0\t10.0000\n 4\t 5.0\t 0\t23.7500\n 5\t10.0\t 8\t30.0000\n 6\t15.0\t 24\t30.0000 <<\n 7\t20.0\t 36\t31.2500\n 8\t50.0\t 64\t38.7500\n 9\t100.0\t 69\t45.6250\n 10\t200.0\t 73\t61.8750\n 11\t400.0\t 90\t92.5000\n 12\t800.0\t 84\t148.7500\n\nstep 50000 | λ index 10 | wl_increment = 0.625 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 36\t0.0000\n 1\t 0.5\t 41\t-0.6250\n 2\t 1.0\t 47\t-0.6250\n 3\t 2.0\t 50\t1.2500\n 4\t 5.0\t 61\t8.1250\n 5\t10.0\t 68\t15.0000\n 6\t15.0\t 81\t16.8750\n 7\t20.0\t 89\t20.6250\n 8\t50.0\t 126\t22.5000\n 9\t100.0\t 140\t23.7500\n 10\t200.0\t 168\t25.0000 <<\n 11\t400.0\t 231\t26.8750\n 12\t800.0\t 310\t30.0000\n\nstep 60000 | λ index 12 | wl_increment = 0.625 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 142\t0.0000\n 1\t 0.5\t 134\t7.5000\n 2\t 1.0\t 125\t16.8750\n 3\t 2.0\t 124\t21.2500\n 4\t 5.0\t 124\t35.0000\n 5\t10.0\t 140\t36.2500\n 6\t15.0\t 154\t37.5000\n 7\t20.0\t 169\t36.8750\n 8\t50.0\t 207\t38.1250\n 9\t100.0\t 215\t43.1250\n 10\t200.0\t 240\t46.2500\n 11\t400.0\t 300\t50.0000\n 12\t800.0\t 374\t56.2500 <<\n\nstep 70000 | λ index 1 | wl_increment = 0.625 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 195\t0.0000\n 1\t 0.5\t 194\t3.1250 <<\n 2\t 1.0\t 195\t6.2500\n 3\t 2.0\t 196\t9.3750\n 4\t 5.0\t 212\t13.1250\n 5\t10.0\t 225\t16.2500\n 6\t15.0\t 239\t17.5000\n 7\t20.0\t 248\t20.6250\n 8\t50.0\t 286\t21.8750\n 9\t100.0\t 303\t21.2500\n 10\t200.0\t 329\t23.7500\n 11\t400.0\t 383\t31.2500\n 12\t800.0\t 443\t46.2500\n\nstep 80000 | λ index 9 | wl_increment = 0.625 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 266\t0.0000\n 1\t 0.5\t 268\t1.2500\n 2\t 1.0\t 272\t2.5000\n 3\t 2.0\t 271\t6.8750\n 4\t 5.0\t 288\t10.0000\n 5\t10.0\t 301\t13.1250\n 6\t15.0\t 320\t11.2500\n 7\t20.0\t 332\t12.5000\n 8\t50.0\t 363\t18.1250\n 9\t100.0\t 374\t21.2500 <<\n 10\t200.0\t 400\t23.7500\n 11\t400.0\t 457\t29.3750\n 12\t800.0\t 536\t32.5000\n\nstep 90000 | λ index 1 | wl_increment = 0.625 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 367\t0.0000\n 1\t 0.5\t 370\t0.6250 <<\n 2\t 1.0\t 373\t2.5000\n 3\t 2.0\t 368\t9.3750\n 4\t 5.0\t 363\t26.2500\n 5\t10.0\t 373\t31.2500\n 6\t15.0\t 387\t32.5000\n 7\t20.0\t 399\t33.7500\n 8\t50.0\t 434\t36.8750\n 9\t100.0\t 441\t42.5000\n 10\t200.0\t 466\t45.6250\n 11\t400.0\t 519\t53.7500\n 12\t800.0\t 588\t63.1250\n\nstep 100000 | λ index 3 | wl_increment = 0.625 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 421\t0.0000\n 1\t 0.5\t 416\t5.6250\n 2\t 1.0\t 426\t3.1250\n 3\t 2.0\t 431\t3.7500 <<\n 4\t 5.0\t 448\t6.8750\n 5\t10.0\t 460\t10.6250\n 6\t15.0\t 469\t15.0000\n 7\t20.0\t 477\t18.7500\n 8\t50.0\t 514\t20.6250\n 9\t100.0\t 528\t21.8750\n 10\t200.0\t 554\t24.3750\n 11\t400.0\t 615\t27.5000\n 12\t800.0\t 689\t33.7500\n\nstep 110000 | λ index 7 | wl_increment = 0.625 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 512\t0.0000\n 1\t 0.5\t 512\t2.5000\n 2\t 1.0\t 511\t6.8750\n 3\t 2.0\t 510\t11.2500\n 4\t 5.0\t 523\t16.8750\n 5\t10.0\t 536\t20.0000\n 6\t15.0\t 549\t21.8750\n 7\t20.0\t 560\t23.7500 <<\n 8\t50.0\t 591\t29.3750\n 9\t100.0\t 602\t32.5000\n 10\t200.0\t 617\t41.8750\n 11\t400.0\t 675\t46.8750\n 12\t800.0\t 750\t52.5000\n\nstep 120000 | λ index 11 | wl_increment = 0.625 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 577\t0.0000\n 1\t 0.5\t 577\t2.5000\n 2\t 1.0\t 581\t3.7500\n 3\t 2.0\t 580\t8.1250\n 4\t 5.0\t 597\t11.2500\n 5\t10.0\t 611\t13.7500\n 6\t15.0\t 625\t15.0000\n 7\t20.0\t 636\t16.8750\n 8\t50.0\t 668\t21.8750\n 9\t100.0\t 678\t25.6250\n 10\t200.0\t 709\t25.0000\n 11\t400.0\t 769\t28.7500 <<\n 12\t800.0\t 840\t36.8750\n\nstep 130000 | λ index 8 | wl_increment = 0.625 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 667\t0.0000\n 1\t 0.5\t 668\t1.8750\n 2\t 1.0\t 671\t3.7500\n 3\t 2.0\t 664\t11.8750\n 4\t 5.0\t 666\t24.3750\n 5\t10.0\t 682\t25.6250\n 6\t15.0\t 697\t26.2500\n 7\t20.0\t 714\t24.3750\n 8\t50.0\t 750\t26.8750 <<\n 9\t100.0\t 759\t31.2500\n 10\t200.0\t 773\t41.2500\n 11\t400.0\t 828\t48.1250\n 12\t800.0\t 909\t50.0000\n\nstep 140000 | λ index 5 | wl_increment = 0.3125 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 0\t0.0000\n 1\t 0.5\t 0\t3.1250\n 2\t 1.0\t 0\t7.5000\n 3\t 2.0\t 0\t15.6250\n 4\t 5.0\t 32\t21.2500\n 5\t10.0\t 45\t23.4375 <<\n 6\t15.0\t 49\t24.0625\n 7\t20.0\t 51\t25.3125\n 8\t50.0\t 55\t27.1875\n 9\t100.0\t 55\t29.0625\n 10\t200.0\t 73\t31.5625\n 11\t400.0\t 111\t35.3125\n 12\t800.0\t 129\t40.9375\n\nstep 150000 | λ index 3 | wl_increment = 0.3125 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 43\t0.0000\n 1\t 0.5\t 47\t1.8750\n 2\t 1.0\t 60\t2.1875\n 3\t 2.0\t 80\t4.0625 <<\n 4\t 5.0\t 120\t7.1875\n 5\t10.0\t 132\t9.6875\n 6\t15.0\t 129\t12.5000\n 7\t20.0\t 130\t14.0625\n 8\t50.0\t 135\t15.6250\n 9\t100.0\t 139\t16.2500\n 10\t200.0\t 158\t18.4375\n 11\t400.0\t 201\t20.6250\n 12\t800.0\t 226\t24.0625\n\nstep 160000 | λ index 9 | wl_increment = 0.3125 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 122\t0.0000\n 1\t 0.5\t 126\t1.8750\n 2\t 1.0\t 136\t3.1250\n 3\t 2.0\t 160\t3.7500\n 4\t 5.0\t 203\t5.9375\n 5\t10.0\t 212\t9.3750\n 6\t15.0\t 217\t9.6875\n 7\t20.0\t 219\t10.9375\n 8\t50.0\t 213\t15.9375\n 9\t100.0\t 206\t20.0000 <<\n 10\t200.0\t 228\t21.2500\n 11\t400.0\t 268\t24.3750\n 12\t800.0\t 290\t28.7500\n\n" ], [ "### compare result with experiment\n\nee_temp = EESampler_RigidThreeParticle()\n\ntheory_dG_in_kT = ee_temp.theory_dg_in_kT()\nprint('theory_dG_in_kT', theory_dG_in_kT)\n\nprint('ee.g',ee.g)\nplt.figure(figsize=(10,4))\n\n# Plot the final free energy estimates as a function of k\nplt.subplot(1,2,1)\nplt.plot(ee.k_values, ee.g, 'o-', label='EE result')\nplt.plot(ee.k_values[1:], theory_dG_in_kT, 'o-', label='theory')\nplt.xlabel('$k$ (kJ/nm$^2$)')\nplt.ylabel('$\\Delta G_{rest}$ (kT)')\n#lt.xscale('log')\nplt.legend(loc='best')\n\n# Plot the convergence of the free energy\nstep = traj.loc[:,'step'].values\nfree_energy = traj.loc[:,'free_energy'].values\nplt.subplot(1,2,2)\nplt.plot(step, free_energy, 'r-')\nplt.xlabel('step')\nplt.ylabel('$\\Delta G_{rest}$ (kT)')\n# plt.legend(loc='best')\n\n\n", "self.d 1.0\nself.c 1.0\nself.e 2.0\nself.d 1.0\nself.c 1.0\nself.e 2.0\nself.L 8.0\nkc_coeff 5.0\nkp1_coeff 6.0\ntheory_dG_in_kT [ 3.63910456 5.7185461 7.79798765 10.54685984 12.62630138 13.84269671\n 14.70574292 17.45461512 19.53405666 21.6134982 23.69293975 25.77238129]\nee.g [ 0. 3.6289978 5.2804184 7.11612701 9.68471527 11.73103333\n 12.99034119 13.8904953 16.87911987 19.1463089 21.32175446 23.47545624\n 25.64460754]\n" ] ], [ [ "### Let's try parameters from an actual simulation: REF_RL/RUN0\n", "_____no_output_____" ] ], [ [ "\"\"\"\n[tud67309@login2 RUN0]$ cat LIG_res.itp\n;LIG_res.itp\n[ position_restraints ]\n;i funct fcx fcy fcz\n6 1 800 800 800\n35 1 800 800 800\n23 1 800 800 800\n\"\"\"\n\n# From LIG_h.pdb:\n\"\"\"\nATOM 6 C28 LIG A 1 11.765 -16.536 1.909 1.00 0.00 C \n...\nATOM 23 C23 LIG A 1 12.358 -10.647 7.766 1.00 0.00 C \n...\nATOM 35 C17 LIG A 1 20.883 -7.674 2.314 1.00 0.00 C \n\"\"\"\n\nx0 = np.zeros((3,3))\nx0[0,:] = np.array([1.1765, -1.6536, 0.1909]) # converted to nm\nx0[1,:] = np.array([1.2358, -1.0647, 0.7766]) # converted to nm\nx0[2,:] = np.array([2.0883, -0.7674, 0.2314]) # converted to nm\n\nV0 = 1660.0 # in Å^3 is the standard volume\nL0 = ((1660.0)**(1/3)) * 0.1 # converted to nm\nprint('L0', L0, 'nm')\n\n\nee_REF0 = EESampler_RigidThreeParticle(L=L0, x0=x0, a1=x0)\n\ntheory_dG_in_kT = ee_REF0.theory_dg_in_kT()\nprint('theory_dG_in_kT', theory_dG_in_kT)\n\nprint('ee_REF0.g', ee_REF0.g)\nplt.figure(figsize=(10,4))\n\n# Plot the final free energy estimates as a function of k\nplt.subplot(1,2,1)\n#plt.plot(ee.k_values, ee.g, 'o-', label='EE result')\nplt.plot(ee_REF0.k_values[1:], theory_dG_in_kT, 'o-', label='theory')\nplt.xlabel('$k$ (kJ/nm$^2$)')\nplt.ylabel('$\\Delta G_{rest}$ (kT)')\nplt.legend(loc='best')", "L0 1.1840481475428983 nm\nself.d 0.8326849284092993\nself.c 1.0486785581066906\nself.e 0.720168877255378\nself.d 0.8326849284092993\nself.c 1.0486785581066906\nself.e 0.720168877255378\nself.L 1.1840481475428983\nkc_coeff 1.7480098038626308\nkp1_coeff 3.334083521100217\ntheory_dG_in_kT [-2.88375874 -0.80431719 1.27512435 4.02399654 6.10343808 7.31983341\n 8.18287963 10.93175182 13.01119336 15.09063491 17.17007645 19.24951799]\nee_REF0.g [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n" ] ], [ [ "### REF_RL/RUN1", "_____no_output_____" ] ], [ [ "\"\"\"\n$ cat LIG_res.itp\n;LIG_res.itp\n[ position_restraints ]\n;i funct fcx fcy fcz\n11 1 800 800 800\n25 1 800 800 800\n2 1 800 800 800\n\"\"\"\n\n# From LIG_h.pdb:\n\"\"\"\nATOM 2 C12 LIG A 1 6.050 0.774 17.871 1.00 0.00 C \nATOM 11 C15 LIG A 1 2.770 2.355 21.054 1.00 0.00 C \nATOM 25 C4 LIG A 1 13.466 -1.210 22.191 1.00 0.00 C \n\"\"\"\n\nx0 = np.zeros((3,3))\nx0[0,:] = np.array([0.6050, 0.0774, 1.7871]) # converted to nm\nx0[1,:] = np.array([0.2770, 0.2355, 2.1054]) # converted to nm\nx0[2,:] = np.array([1.3466, -0.1210, 2.2191]) # converted to nm\n\nV0 = 1660.0 # in Å^3 is the standard volume\nL0 = ((1660.0)**(1/3)) * 0.1 # converted to nm\nprint('L0', L0, 'nm')\n\n\nee_REF1 = EESampler_RigidThreeParticle(L=L0, x0=x0, a1=x0)\n\n# The *theory* for the triple-restraint rigid rotor says that \n# dG/kT = -ln ( [(2*\\pi)/(\\beta *3k )]^{3/2} / L^3\n# [(2*\\pi)/(\\beta*2k))]^{1/2} * [(2*\\pi)/(\\beta *(2+c^2/(d/2)^2)k )]^{1/2} / 4 \\pi (d/2)^2\n# [(2*\\pi)/(\\beta *k )]^{1/2} / 2 \\pi c\n\nprint('ee_REF1.d', ee_REF1.d)\nprint('ee_REF1.c', ee_REF1.c)\nprint('ee_REF1.L', ee_REF1.L)\n\nk_prime_coeff = 2.0 + (ee.c/(ee.d/2.))**2.0\nprint('k_prime_coeff', k_prime_coeff)\n\ntheory_dG_in_kT = -1.0*( 3.0/2.0*np.log(2.0*np.pi*ee_REF1.RT/(3.0*ee_REF1.k_values[1:])) \\\n + 1.0/2.0*np.log(2.0*np.pi*ee_REF1.RT/(2.0*ee_REF1.k_values[1:])) \\\n + 1.0/2.0*np.log(2.0*np.pi*ee_REF1.RT/(k_prime_coeff*ee_REF0.k_values[1:])) \\\n + 1.0/2.0*np.log(2.0*np.pi*ee_REF1.RT/ee_REF1.k_values[1:]) \\\n - np.log( ee_REF1.L**3 * 8.0 * (np.pi**2) * (ee_REF1.d/2.0)**2 * ee_REF1.c ) )\n\nprint('theory_dG_in_kT', theory_dG_in_kT)\n\nprint('ee_REF1.g', ee_REF1.g)\nplt.figure(figsize=(10,4))\n\n# Plot the final free energy estimates as a function of k\nplt.subplot(1,2,1)\n#plt.plot(ee.k_values, ee.g, 'o-', label='EE result')\nplt.plot(ee_REF1.k_values[1:], theory_dG_in_kT, 'o-', label='theory')\nplt.xlabel('$k$ (kJ/nm$^2$)')\nplt.ylabel('$\\Delta G_{rest}$ (kT)')\nplt.legend(loc='best')", "L0 1.1840481475428983 nm\nself.d 0.4836264053998706\nself.c 0.834018605392616\nee_REF1.d 0.4836264053998706\nee_REF1.c 0.834018605392616\nee_REF1.L 1.1840481475428983\nk_prime_coeff 5.999999999996005\ntheory_dG_in_kT [-5.57122688 -3.49178533 -1.41234379 1.3365284 3.41596994 4.63236527\n 5.49541149 8.24428368 10.32372522 12.40316677 14.48260831 16.56204985]\nee_REF1.g [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n" ] ], [ [ "### REF_RL/RUN2", "_____no_output_____" ] ], [ [ "\"\"\";LIG_res.itp\n[ position_restraints ]\n;i funct fcx fcy fcz\n13 1 800 800 800\n19 1 800 800 800\n9 1 800 800 800\n\"\"\"\n\n# From LIG_h.pdb:\n\"\"\"\nATOM 9 C2 LIG A 1 12.189 0.731 23.852 1.00 0.00 C \nATOM 13 CL LIG A 1 14.006 -1.527 21.119 1.00 0.00 Cl \nATOM 19 C13 LIG A 1 3.244 2.176 20.610 1.00 0.00 C \n\"\"\"\n\nx0 = np.zeros((3,3))\nx0[0,:] = 0.1*np.array([12.189, 0.731, 23.852]) # converted to nm\nx0[1,:] = 0.1*np.array([14.006, -1.527, 21.119]) # converted to nm\nx0[2,:] = 0.1*np.array([ 3.244, 2.176, 20.610]) # converted to nm\n\nV0 = 1660.0 # in Å^3 is the standard volume\nL0 = ((1660.0)**(1/3)) * 0.1 # converted to nm\nprint('L0', L0, 'nm')\n\n\nee_REF2 = EESampler_RigidThreeParticle(L=L0, x0=x0, a1=x0)\n\n# The *theory* for the triple-restraint rigid rotor says that \n# dG/kT = -ln ( [(2*\\pi)/(\\beta *3k )]^{3/2} / L^3\n# [(2*\\pi)/(\\beta*2k))]^{1/2} * [(2*\\pi)/(\\beta *(2+c^2/(d/2)^2)k )]^{1/2} / 4 \\pi (d/2)^2\n# [(2*\\pi)/(\\beta *k )]^{1/2} / 2 \\pi c\n\nprint('ee_REF1.d', ee_REF2.d)\nprint('ee_REF1.c', ee_REF2.c)\nprint('ee_REF1.L', ee_REF2.L)\n\nk_prime_coeff = 2.0 + (ee.c/(ee.d/2.))**2.0\nprint('k_prime_coeff', k_prime_coeff)\n\ntheory_dG_in_kT = -1.0*( 3.0/2.0*np.log(2.0*np.pi*ee_REF2.RT/(3.0*ee_REF2.k_values[1:])) \\\n + 1.0/2.0*np.log(2.0*np.pi*ee_REF2.RT/(2.0*ee_REF2.k_values[1:])) \\\n + 1.0/2.0*np.log(2.0*np.pi*ee_REF2.RT/(k_prime_coeff*ee_REF0.k_values[1:])) \\\n + 1.0/2.0*np.log(2.0*np.pi*ee_REF2.RT/ee_REF2.k_values[1:]) \\\n - np.log( ee_REF2.L**3 * 8.0 * (np.pi**2) * (ee_REF2.d/2.0)**2 * ee_REF2.c ) )\n\nprint('theory_dG_in_kT', theory_dG_in_kT)\n\nprint('ee_REF2.g', ee_REF2.g)\nplt.figure(figsize=(10,4))\n\n# Plot the final free energy estimates as a function of k\nplt.subplot(1,2,1)\n#plt.plot(ee.k_values, ee.g, 'o-', label='EE result')\nplt.plot(ee_REF2.k_values[1:], theory_dG_in_kT, 'o-', label='theory')\nplt.xlabel('$k$ (kJ/nm$^2$)')\nplt.ylabel('$\\Delta G_{rest}$ (kT)')\nplt.legend(loc='best')", "L0 1.1840481475428983 nm\nself.d 0.3983634270361678\nself.c 0.9244294074859049\nee_REF1.d 0.3983634270361678\nee_REF1.c 0.9244294074859049\nee_REF1.L 1.1840481475428983\nk_prime_coeff 5.999999999996005\ntheory_dG_in_kT [-5.85620189 -3.77676035 -1.69731881 1.05155339 3.13099493 4.34739025\n 5.21043647 7.95930867 10.03875021 12.11819175 14.19763329 16.27707483]\nee_REF2.g [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n" ], [ "nsteps = 100000\ntraj = ee.sample(nsteps)\n# print(traj['dhdl'])\n\nstep = traj.loc[:,'step'].values\np1 = np.zeros((step.shape[0], 3))\np1[:,0] = traj.loc[:,'x1'].values\np1[:,1] = traj.loc[:,'y1'].values\np1[:,2] = traj.loc[:,'z1'].values\np2 = np.zeros((step.shape[0], 3))\np2[:,0] = traj.loc[:,'x2'].values\np2[:,1] = traj.loc[:,'y2'].values\np2[:,2] = traj.loc[:,'z2'].values\nd = traj.loc[:,'distance'].values\nc = traj.loc[:,'height'].values\n\nimport matplotlib\nfrom matplotlib import pyplot as plt\nplt.figure(figsize=(10,4))\nplt.subplot(1,3,1)\nplt.plot(step, p1)\nplt.plot(step, p2)\nplt.subplot(1,3,2)\nplt.plot(step, d)\nplt.subplot(1,3,3)\nplt.plot(step, c)", "step 0 | λ index 0 | wl_increment = 5.0 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 0\t0.0000 <<\n 1\t 0.5\t 0\t0.0000\n 2\t 1.0\t 0\t0.0000\n 3\t 2.0\t 0\t0.0000\n 4\t 5.0\t 0\t0.0000\n 5\t10.0\t 0\t0.0000\n 6\t15.0\t 0\t0.0000\n 7\t20.0\t 0\t0.0000\n 8\t50.0\t 0\t0.0000\n 9\t100.0\t 0\t0.0000\n 10\t200.0\t 0\t0.0000\n 11\t400.0\t 0\t0.0000\n 12\t800.0\t 0\t0.0000\n\nstep 10000 | λ index 2 | wl_increment = 5.0 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 106\t0.0000\n 1\t 0.5\t 106\t0.0000\n 2\t 1.0\t 107\t-5.0000 <<\n 3\t 2.0\t 107\t-5.0000\n 4\t 5.0\t 103\t15.0000\n 5\t10.0\t 99\t35.0000\n 6\t15.0\t 98\t40.0000\n 7\t20.0\t 95\t55.0000\n 8\t50.0\t 85\t105.0000\n 9\t100.0\t 67\t195.0000\n 10\t200.0\t 27\t395.0000\n 11\t400.0\t 0\t530.0000\n 12\t800.0\t 0\t530.0000\n\nstep 20000 | λ index 8 | wl_increment = 5.0 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 192\t0.0000\n 1\t 0.5\t 192\t0.0000\n 2\t 1.0\t 190\t10.0000\n 3\t 2.0\t 189\t15.0000\n 4\t 5.0\t 188\t20.0000\n 5\t10.0\t 187\t25.0000\n 6\t15.0\t 183\t45.0000\n 7\t20.0\t 180\t60.0000\n 8\t50.0\t 168\t120.0000 <<\n 9\t100.0\t 154\t190.0000\n 10\t200.0\t 122\t350.0000\n 11\t400.0\t 55\t685.0000\n 12\t800.0\t 0\t960.0000\n\nstep 30000 | λ index 2 | wl_increment = 5.0 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 276\t0.0000\n 1\t 0.5\t 277\t-5.0000\n 2\t 1.0\t 276\t0.0000 <<\n 3\t 2.0\t 276\t0.0000\n 4\t 5.0\t 273\t15.0000\n 5\t10.0\t 271\t25.0000\n 6\t15.0\t 271\t25.0000\n 7\t20.0\t 269\t35.0000\n 8\t50.0\t 251\t125.0000\n 9\t100.0\t 233\t215.0000\n 10\t200.0\t 198\t390.0000\n 11\t400.0\t 129\t735.0000\n 12\t800.0\t 0\t1380.0000\n\nstep 40000 | λ index 3 | wl_increment = 5.0 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 359\t0.0000\n 1\t 0.5\t 359\t0.0000\n 2\t 1.0\t 359\t0.0000\n 3\t 2.0\t 359\t0.0000 <<\n 4\t 5.0\t 358\t5.0000\n 5\t10.0\t 354\t25.0000\n 6\t15.0\t 349\t50.0000\n 7\t20.0\t 345\t70.0000\n 8\t50.0\t 329\t150.0000\n 9\t100.0\t 307\t260.0000\n 10\t200.0\t 270\t445.0000\n 11\t400.0\t 198\t805.0000\n 12\t800.0\t 54\t1525.0000\n\nstep 50000 | λ index 2 | wl_increment = 5.0 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 441\t0.0000\n 1\t 0.5\t 441\t0.0000\n 2\t 1.0\t 438\t15.0000 <<\n 3\t 2.0\t 437\t20.0000\n 4\t 5.0\t 435\t30.0000\n 5\t10.0\t 431\t50.0000\n 6\t15.0\t 426\t75.0000\n 7\t20.0\t 423\t90.0000\n 8\t50.0\t 405\t180.0000\n 9\t100.0\t 379\t310.0000\n 10\t200.0\t 335\t530.0000\n 11\t400.0\t 268\t865.0000\n 12\t800.0\t 141\t1500.0000\n\nstep 60000 | λ index 2 | wl_increment = 5.0 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 503\t0.0000\n 1\t 0.5\t 503\t0.0000\n 2\t 1.0\t 503\t0.0000 <<\n 3\t 2.0\t 506\t-15.0000\n 4\t 5.0\t 504\t-5.0000\n 5\t10.0\t 503\t0.0000\n 6\t15.0\t 501\t10.0000\n 7\t20.0\t 500\t15.0000\n 8\t50.0\t 488\t75.0000\n 9\t100.0\t 466\t185.0000\n 10\t200.0\t 430\t365.0000\n 11\t400.0\t 363\t700.0000\n 12\t800.0\t 230\t1365.0000\n\nstep 70000 | λ index 10 | wl_increment = 5.0 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 590\t0.0000\n 1\t 0.5\t 590\t0.0000\n 2\t 1.0\t 590\t0.0000\n 3\t 2.0\t 588\t10.0000\n 4\t 5.0\t 583\t35.0000\n 5\t10.0\t 582\t40.0000\n 6\t15.0\t 580\t50.0000\n 7\t20.0\t 578\t60.0000\n 8\t50.0\t 567\t115.0000\n 9\t100.0\t 548\t210.0000\n 10\t200.0\t 507\t415.0000 <<\n 11\t400.0\t 426\t820.0000\n 12\t800.0\t 271\t1595.0000\n\nstep 80000 | λ index 12 | wl_increment = 5.0 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 665\t0.0000\n 1\t 0.5\t 666\t-5.0000\n 2\t 1.0\t 666\t-5.0000\n 3\t 2.0\t 663\t10.0000\n 4\t 5.0\t 661\t20.0000\n 5\t10.0\t 659\t30.0000\n 6\t15.0\t 657\t40.0000\n 7\t20.0\t 652\t65.0000\n 8\t50.0\t 640\t125.0000\n 9\t100.0\t 621\t220.0000\n 10\t200.0\t 586\t395.0000\n 11\t400.0\t 510\t775.0000\n 12\t800.0\t 354\t1555.0000 <<\n\nstep 90000 | λ index 0 | wl_increment = 5.0 kT\n# ensemble\tk_value\thistogram\tg (kT)\n 0\t 0.0\t 750\t0.0000 <<\n 1\t 0.5\t 749\t5.0000\n 2\t 1.0\t 750\t0.0000\n 3\t 2.0\t 749\t5.0000\n 4\t 5.0\t 744\t30.0000\n 5\t10.0\t 743\t35.0000\n 6\t15.0\t 740\t50.0000\n 7\t20.0\t 739\t55.0000\n 8\t50.0\t 724\t130.0000\n 9\t100.0\t 698\t260.0000\n 10\t200.0\t 650\t500.0000\n 11\t400.0\t 561\t945.0000\n 12\t800.0\t 403\t1735.0000\n\n24970\n" ], [ "# The *theory* for the triple-restraint rigid rotor says that \n# dG/kT = -ln ( [(2*\\pi)/(\\beta *3k )]^{3/2} / L^3\n# [(2*\\pi)/(\\beta*2k))]^{1/2} * [(2*\\pi)/(\\beta *(2+c^2/(d/2)^2)k )]^{1/2} / 4 \\pi (d/2)^2\n# [(2*\\pi)/(\\beta *k )]^{1/2} / 2 \\pi c\n\nprint('ee.d', ee.d)\nprint('ee.c', ee.c)\nprint('ee.L', ee.L)\n\nk_prime_coeff = 2.0 + (ee.c/(ee.d/2.))**2.0\nprint('k_prime_coeff', k_prime_coeff)\n\ntheory_dG_in_kT = -1.0*( 3.0/2.0*np.log(2.0*np.pi*ee.RT/(3.0*ee.k_values[1:])) \\\n + 1.0/2.0*np.log(2.0*np.pi*ee.RT/(2.0*ee.k_values[1:])) \\\n + 1.0/2.0*np.log(2.0*np.pi*ee.RT/(k_prime_coeff*ee.k_values[1:])) \\\n + 1.0/2.0*np.log(2.0*np.pi*ee.RT/ee.k_values[1:]) \\\n - np.log( ee.L**3 * 8.0 * (np.pi**2) * (ee.d/2.0)**2 * ee.c ) )\n\nprint('theory_dG_in_kT', theory_dG_in_kT)\n\nprint('ee.g',ee.g)\nplt.figure(figsize=(10,4))\n\n# Plot the final free energy estimates as a function of k\nplt.subplot(1,2,1)\n#plt.plot(ee.k_values, ee.g, 'o-', label='EE result')\nplt.plot(ee.k_values[1:], theory_dG_in_kT, 'o-', label='theory')\nplt.xlabel('$k$ (kJ/nm$^2$)')\nplt.ylabel('$\\Delta G_{rest}$ (kT)')\nplt.legend(loc='best')\n\n# Plot the convergence of the free energy\nstep = traj.loc[:,'step'].values\nfree_energy = traj.loc[:,'free_energy'].values\nplt.subplot(1,2,2)\nplt.plot(step, free_energy, 'r-')\nplt.xlabel('step')\nplt.ylabel('$\\Delta G_{rest}$ (kT)')\n# plt.legend(loc='best')\n\n\n", "ee.d 0.5844506200868992\nee.c 0.40967777340459033\nee.L 1.1840481475428983\nk_prime_coeff 3.965391840602323\ntheory_dG_in_kT [-6.1104699 -4.03102836 -1.95158682 0.79728538 2.87672692 4.09312225\n 4.95616846 7.70504066 9.7844822 11.86392374 13.94336528 16.02280683]\nee.g [ 0. 5. 5. 20. 35. 70. 75. 95. 200. 330. 605. 1045.\n 1725.]\n" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "from IPython.display import Image\nfrom IPython.core.display import HTML \nfrom sympy import *\nImage(url= \"https://i.imgur.com/mlreXuN.png\")", "_____no_output_____" ], [ "Image(url= \"https://i.imgur.com/ZspIgwo.png\")\n#https://www.albert.io/blog/how-to-use-the-midpoint-rule-in-ap-calculus/", "_____no_output_____" ], [ "x,a,b,n = symbols('x a b n')\n\nf = lambda x: 25- x**2\na = 0\nb = 5\nn = 5\ndx = (b-a)/n\nmylist = []\n\n#right hand sum\nfor i in range(1,6):\n g = f(i)* dx\n\n print(g)\n print('i =',i)\n print(' ')\n mylist.append(g)\n\nprint(\"total = \",sum(mylist[0:10]))", "24.0\ni = 1\n \n21.0\ni = 2\n \n16.0\ni = 3\n \n9.0\ni = 4\n \n0.0\ni = 5\n \ntotal = 70.0\n" ], [ "x,a,b,n = symbols('x a b n')\n\nf = lambda x: 25- x**2\na = 0\nb = 5\nn = 5\ndx = (b-a)/n\nmylist = []\n\n#left hand sum\nfor i in range(0,5):\n g = f(i)* dx\n\n print(g)\n print('i =',i)\n print(' ')\n mylist.append(g)\n\nprint(\"total = \",sum(mylist[0:10]))", "25.0\ni = 0\n \n24.0\ni = 1\n \n21.0\ni = 2\n \n16.0\ni = 3\n \n9.0\ni = 4\n \ntotal = 95.0\n" ], [ "plot(f(x))", "_____no_output_____" ], [ "Image(url= \"https://i.imgur.com/UZzymrR.png\")", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Creating a Sentiment Analysis Web App\n## Using PyTorch and SageMaker\n\n_Deep Learning Nanodegree Program | Deployment_\n\n---\n\nNow that we have a basic understanding of how SageMaker works we will try to use it to construct a complete project from end to end. Our goal will be to have a simple web page which a user can use to enter a movie review. The web page will then send the review off to our deployed model which will predict the sentiment of the entered review.\n\n## Instructions\n\nSome template code has already been provided for you, and you will need to implement additional functionality to successfully complete this notebook. You will not need to modify the included code beyond what is requested. Sections that begin with '**TODO**' in the header indicate that you need to complete or implement some portion within them. Instructions will be provided for each section and the specifics of the implementation are marked in the code block with a `# TODO: ...` comment. Please be sure to read the instructions carefully!\n\nIn addition to implementing code, there will be questions for you to answer which relate to the task and your implementation. Each section where you will answer a question is preceded by a '**Question:**' header. Carefully read each question and provide your answer below the '**Answer:**' header by editing the Markdown cell.\n\n> **Note**: Code and Markdown cells can be executed using the **Shift+Enter** keyboard shortcut. In addition, a cell can be edited by typically clicking it (double-click for Markdown cells) or by pressing **Enter** while it is highlighted.\n\n## General Outline\n\nRecall the general outline for SageMaker projects using a notebook instance.\n\n1. Download or otherwise retrieve the data.\n2. Process / Prepare the data.\n3. Upload the processed data to S3.\n4. Train a chosen model.\n5. Test the trained model (typically using a batch transform job).\n6. Deploy the trained model.\n7. Use the deployed model.\n\nFor this project, you will be following the steps in the general outline with some modifications. \n\nFirst, you will not be testing the model in its own step. You will still be testing the model, however, you will do it by deploying your model and then using the deployed model by sending the test data to it. One of the reasons for doing this is so that you can make sure that your deployed model is working correctly before moving forward.\n\nIn addition, you will deploy and use your trained model a second time. In the second iteration you will customize the way that your trained model is deployed by including some of your own code. In addition, your newly deployed model will be used in the sentiment analysis web app.", "_____no_output_____" ], [ "## Step 1: Downloading the data\n\nAs in the XGBoost in SageMaker notebook, we will be using the [IMDb dataset](http://ai.stanford.edu/~amaas/data/sentiment/)\n\n> Maas, Andrew L., et al. [Learning Word Vectors for Sentiment Analysis](http://ai.stanford.edu/~amaas/data/sentiment/). In _Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies_. Association for Computational Linguistics, 2011.", "_____no_output_____" ] ], [ [ "%mkdir ../data\n!wget -O ../data/aclImdb_v1.tar.gz http://ai.stanford.edu/~amaas/data/sentiment/aclImdb_v1.tar.gz\n!tar -zxf ../data/aclImdb_v1.tar.gz -C ../data", "mkdir: cannot create directory ‘../data’: File exists\n--2020-10-07 18:13:58-- http://ai.stanford.edu/~amaas/data/sentiment/aclImdb_v1.tar.gz\nResolving ai.stanford.edu (ai.stanford.edu)... 171.64.68.10\nConnecting to ai.stanford.edu (ai.stanford.edu)|171.64.68.10|:80... connected.\nHTTP request sent, awaiting response... 200 OK\nLength: 84125825 (80M) [application/x-gzip]\nSaving to: ‘../data/aclImdb_v1.tar.gz’\n\n../data/aclImdb_v1. 100%[===================>] 80.23M 9.44MB/s in 10s \n\n2020-10-07 18:14:08 (7.99 MB/s) - ‘../data/aclImdb_v1.tar.gz’ saved [84125825/84125825]\n\n" ] ], [ [ "## Step 2: Preparing and Processing the data\n\nAlso, as in the XGBoost notebook, we will be doing some initial data processing. The first few steps are the same as in the XGBoost example. To begin with, we will read in each of the reviews and combine them into a single input structure. Then, we will split the dataset into a training set and a testing set.", "_____no_output_____" ] ], [ [ "import os\nimport glob\n\ndef read_imdb_data(data_dir='../data/aclImdb'):\n data = {}\n labels = {}\n \n for data_type in ['train', 'test']:\n data[data_type] = {}\n labels[data_type] = {}\n \n for sentiment in ['pos', 'neg']:\n data[data_type][sentiment] = []\n labels[data_type][sentiment] = []\n \n path = os.path.join(data_dir, data_type, sentiment, '*.txt')\n files = glob.glob(path)\n \n for f in files:\n with open(f) as review:\n data[data_type][sentiment].append(review.read())\n # Here we represent a positive review by '1' and a negative review by '0'\n labels[data_type][sentiment].append(1 if sentiment == 'pos' else 0)\n \n assert len(data[data_type][sentiment]) == len(labels[data_type][sentiment]), \\\n \"{}/{} data size does not match labels size\".format(data_type, sentiment)\n \n return data, labels", "_____no_output_____" ], [ "data, labels = read_imdb_data()\nprint(\"IMDB reviews: train = {} pos / {} neg, test = {} pos / {} neg\".format(\n len(data['train']['pos']), len(data['train']['neg']),\n len(data['test']['pos']), len(data['test']['neg'])))", "IMDB reviews: train = 12500 pos / 12500 neg, test = 12500 pos / 12500 neg\n" ] ], [ [ "Now that we've read the raw training and testing data from the downloaded dataset, we will combine the positive and negative reviews and shuffle the resulting records.", "_____no_output_____" ] ], [ [ "from sklearn.utils import shuffle\n\ndef prepare_imdb_data(data, labels):\n \"\"\"Prepare training and test sets from IMDb movie reviews.\"\"\"\n \n #Combine positive and negative reviews and labels\n data_train = data['train']['pos'] + data['train']['neg']\n data_test = data['test']['pos'] + data['test']['neg']\n labels_train = labels['train']['pos'] + labels['train']['neg']\n labels_test = labels['test']['pos'] + labels['test']['neg']\n \n #Shuffle reviews and corresponding labels within training and test sets\n data_train, labels_train = shuffle(data_train, labels_train)\n data_test, labels_test = shuffle(data_test, labels_test)\n \n # Return a unified training data, test data, training labels, test labets\n return data_train, data_test, labels_train, labels_test", "_____no_output_____" ], [ "train_X, test_X, train_y, test_y = prepare_imdb_data(data, labels)\nprint(\"IMDb reviews (combined): train = {}, test = {}\".format(len(train_X), len(test_X)))", "IMDb reviews (combined): train = 25000, test = 25000\n" ] ], [ [ "Now that we have our training and testing sets unified and prepared, we should do a quick check and see an example of the data our model will be trained on. This is generally a good idea as it allows you to see how each of the further processing steps affects the reviews and it also ensures that the data has been loaded correctly.", "_____no_output_____" ] ], [ [ "print(train_X[100])\nprint(train_y[100])", "\"The Straight Story\" is a truly beautiful movie about an elderly man named Alvin Straight, who rides his lawnmower across the country to visit his estranged, dying brother. But that's just the basic synapsis...this movie is about so much more than that. This was Richard's Farnworth's last role before he died, and it's definitely one that he will be remembered for. He's a stubborn old man, not unlike a lot of the old men that you and I probably know. <br /><br />\"The Straight Story\" is a movie that everyone should watch at least once in their lives. It will reach down and touch some part of you, at least if you have a heart, it will.\n1\n" ] ], [ [ "The first step in processing the reviews is to make sure that any html tags that appear should be removed. In addition we wish to tokenize our input, that way words such as *entertained* and *entertaining* are considered the same with regard to sentiment analysis.", "_____no_output_____" ] ], [ [ "import nltk\nfrom nltk.corpus import stopwords\nfrom nltk.stem.porter import *\n\nimport re\nfrom bs4 import BeautifulSoup\n\ndef review_to_words(review):\n nltk.download(\"stopwords\", quiet=True)\n stemmer = PorterStemmer()\n \n text = BeautifulSoup(review, \"html.parser\").get_text() # Remove HTML tags\n text = re.sub(r\"[^a-zA-Z0-9]\", \" \", text.lower()) # Convert to lower case\n words = text.split() # Split string into words\n words = [w for w in words if w not in stopwords.words(\"english\")] # Remove stopwords\n words = [PorterStemmer().stem(w) for w in words] # stem\n \n return words", "_____no_output_____" ], [ "import re\n\nREPLACE_NO_SPACE = re.compile(\"(\\.)|(\\;)|(\\:)|(\\!)|(\\')|(\\?)|(\\,)|(\\\")|(\\()|(\\))|(\\[)|(\\])\")\nREPLACE_WITH_SPACE = re.compile(\"(<br\\s*/><br\\s*/>)|(\\-)|(\\/)\")\n\ndef review_to_words_2(review):\n words = REPLACE_NO_SPACE.sub(\"\", review.lower())\n words = REPLACE_WITH_SPACE.sub(\" \", words)\n return words", "_____no_output_____" ] ], [ [ "The `review_to_words` method defined above uses `BeautifulSoup` to remove any html tags that appear and uses the `nltk` package to tokenize the reviews. As a check to ensure we know how everything is working, try applying `review_to_words` to one of the reviews in the training set.", "_____no_output_____" ] ], [ [ "# TODO: Apply review_to_words to a review (train_X[100] or any other review)\nreview_to_words(train_X[100])", "_____no_output_____" ] ], [ [ "**Question:** Above we mentioned that `review_to_words` method removes html formatting and allows us to tokenize the words found in a review, for example, converting *entertained* and *entertaining* into *entertain* so that they are treated as though they are the same word. What else, if anything, does this method do to the input?", "_____no_output_____" ], [ "**Answer:**1)remove any non-alpha numeric characters (like punctuation), which are not useful, to improve the accuracy 2) remove all the Remove all the predefined english stopwords, like is, a, of, the, to, his......which are neutral words,to improve the accuracy 3) convert all in lower cases 4) get the only stem parts of the words", "_____no_output_____" ], [ "The method below applies the `review_to_words` method to each of the reviews in the training and testing datasets. In addition it caches the results. This is because performing this processing step can take a long time. This way if you are unable to complete the notebook in the current session, you can come back without needing to process the data a second time.", "_____no_output_____" ] ], [ [ "import pickle\n\ncache_dir = os.path.join(\"../cache\", \"sentiment_analysis\") # where to store cache files\nos.makedirs(cache_dir, exist_ok=True) # ensure cache directory exists\n\ndef preprocess_data(data_train, data_test, labels_train, labels_test,\n cache_dir=cache_dir, cache_file=\"preprocessed_data.pkl\"):\n \"\"\"Convert each review to words; read from cache if available.\"\"\"\n\n # If cache_file is not None, try to read from it first\n cache_data = None\n if cache_file is not None:\n try:\n with open(os.path.join(cache_dir, cache_file), \"rb\") as f:\n cache_data = pickle.load(f)\n print(\"Read preprocessed data from cache file:\", cache_file)\n except:\n pass # unable to read from cache, but that's okay\n \n # If cache is missing, then do the heavy lifting\n if cache_data is None:\n # Preprocess training and test data to obtain words for each review\n #words_train = list(map(review_to_words, data_train))\n #words_test = list(map(review_to_words, data_test))\n words_train = [review_to_words(review) for review in data_train]\n words_test = [review_to_words(review) for review in data_test]\n \n # Write to cache file for future runs\n if cache_file is not None:\n cache_data = dict(words_train=words_train, words_test=words_test,\n labels_train=labels_train, labels_test=labels_test)\n with open(os.path.join(cache_dir, cache_file), \"wb\") as f:\n pickle.dump(cache_data, f)\n print(\"Wrote preprocessed data to cache file:\", cache_file)\n else:\n # Unpack data loaded from cache file\n words_train, words_test, labels_train, labels_test = (cache_data['words_train'],\n cache_data['words_test'], cache_data['labels_train'], cache_data['labels_test'])\n \n return words_train, words_test, labels_train, labels_test", "_____no_output_____" ], [ "# Preprocess data\ntrain_X, test_X, train_y, test_y = preprocess_data(train_X, test_X, train_y, test_y)", "Read preprocessed data from cache file: preprocessed_data.pkl\n" ] ], [ [ "## Transform the data\n\nIn the XGBoost notebook we transformed the data from its word representation to a bag-of-words feature representation. For the model we are going to construct in this notebook we will construct a feature representation which is very similar. To start, we will represent each word as an integer. Of course, some of the words that appear in the reviews occur very infrequently and so likely don't contain much information for the purposes of sentiment analysis. The way we will deal with this problem is that we will fix the size of our working vocabulary and we will only include the words that appear most frequently. We will then combine all of the infrequent words into a single category and, in our case, we will label it as `1`.\n\nSince we will be using a recurrent neural network, it will be convenient if the length of each review is the same. To do this, we will fix a size for our reviews and then pad short reviews with the category 'no word' (which we will label `0`) and truncate long reviews.", "_____no_output_____" ], [ "### (TODO) Create a word dictionary\n\nTo begin with, we need to construct a way to map words that appear in the reviews to integers. Here we fix the size of our vocabulary (including the 'no word' and 'infrequent' categories) to be `5000` but you may wish to change this to see how it affects the model.\n\n> **TODO:** Complete the implementation for the `build_dict()` method below. Note that even though the vocab_size is set to `5000`, we only want to construct a mapping for the most frequently appearing `4998` words. This is because we want to reserve the special labels `0` for 'no word' and `1` for 'infrequent word'.", "_____no_output_____" ] ], [ [ "import numpy as np\nfrom collections import Counter\ndef build_dict(data, vocab_size = 5000):\n \"\"\"Construct and return a dictionary mapping each of the most frequently appearing words to a unique integer.\"\"\"\n \n # TODO: Determine how often each word appears in `data`. Note that `data` is a list of sentences and that a\n # sentence is a list of words.\n \n word_count ={} # A dict storing the words that appear in the reviews along with how often they occur\n ####Begin: Yanfei First Try#############\n #word_count =Counter(x for xs in data for x in set(xs)) # it is a sorted dictionary already, with the most frequently appearing word in word_count[0], the last frequently appearing word in word_count[-1]\n # TODO: Sort the words found in `data` so that sorted_words[0] is the most frequently appearing word and\n # sorted_words[-1] is the least frequently appearing word.\n #words_sorted=sorted(words.items(), key=lambda x: x[1], reverse=True) # it seems not nesseary, since it is a sorted dictionary already, with the most frequently appearing word first, the last frequently appearing word last\n # sorted_words = None\n # sorted_words = list(word_count.keys()) \n ####End: Yanfei First Try#############\n \n ####Begin: Yanfei Second Try############\n review_words = []\n for review_id in range(len(data)):\n for review_word in data[review_id]:\n review_words.append(review_word)\n word_count = Counter(review_words)\n sorted_words = sorted(word_count, key=word_count.get, reverse=True) # TODO: Sort the words found in `data` so that sorted_words[0] is the most frequently appearing word and\n # sorted_words[-1] is the least frequently appearing word.\n ###End: Yanfei Second Try###############\n \n ####Begin: Yanfei Third Try############\n # word_count =Counter(x for xs in data for x in set(xs))\n # sorted_words = sorted(word_count, key=word_count.get, reverse=True) # TODO: Sort the words found in `data` so that sorted_words[0] is the most frequently appearing word and\n # sorted_words[-1] is the least frequently appearing word.\n ###End: Yanfei Third Try###############\n \n word_dict = {} # This is what we are building, a dictionary that translates words into integers\n for idx, word in enumerate(sorted_words[:vocab_size - 2]): # The -2 is so that we save room for the 'no word'\n word_dict[word] = idx + 2 # 'infrequent' labels\n \n return word_dict", "_____no_output_____" ], [ "word_dict = build_dict(train_X)", "_____no_output_____" ] ], [ [ "**Question:** What are the five most frequently appearing (tokenized) words in the training set? Does it makes sense that these words appear frequently in the training set?", "_____no_output_____" ], [ "**Answer:**['one', 'come', 'cartoon', 'long', 'minut'], yes, it make sense to see it in the movie review.", "_____no_output_____" ] ], [ [ "# TODO: Use this space to determine the five most frequently appearing words in the training set.\n list(word_dict.keys())[0:5] ", "_____no_output_____" ] ], [ [ "### Save `word_dict`\n\nLater on when we construct an endpoint which processes a submitted review we will need to make use of the `word_dict` which we have created. As such, we will save it to a file now for future use.", "_____no_output_____" ] ], [ [ "data_dir = '../data/pytorch' # The folder we will use for storing data\nif not os.path.exists(data_dir): # Make sure that the folder exists\n os.makedirs(data_dir)", "_____no_output_____" ], [ "with open(os.path.join(data_dir, 'word_dict.pkl'), \"wb\") as f:\n pickle.dump(word_dict, f)", "_____no_output_____" ] ], [ [ "### Transform the reviews\n\nNow that we have our word dictionary which allows us to transform the words appearing in the reviews into integers, it is time to make use of it and convert our reviews to their integer sequence representation, making sure to pad or truncate to a fixed length, which in our case is `500`.", "_____no_output_____" ] ], [ [ "def convert_and_pad(word_dict, sentence, pad=500):\n NOWORD = 0 # We will use 0 to represent the 'no word' category\n INFREQ = 1 # and we use 1 to represent the infrequent words, i.e., words not appearing in word_dict\n \n working_sentence = [NOWORD] * pad\n \n for word_index, word in enumerate(sentence[:pad]):\n if word in word_dict:\n working_sentence[word_index] = word_dict[word]\n else:\n working_sentence[word_index] = INFREQ\n \n return working_sentence, min(len(sentence), pad)\n\ndef convert_and_pad_data(word_dict, data, pad=500):\n result = []\n lengths = []\n \n for sentence in data:\n converted, leng = convert_and_pad(word_dict, sentence, pad)\n result.append(converted)\n lengths.append(leng)\n \n return np.array(result), np.array(lengths)", "_____no_output_____" ], [ "train_X, train_X_len = convert_and_pad_data(word_dict, train_X)\ntest_X, test_X_len = convert_and_pad_data(word_dict, test_X)", "_____no_output_____" ] ], [ [ "As a quick check to make sure that things are working as intended, check to see what one of the reviews in the training set looks like after having been processeed. Does this look reasonable? What is the length of a review in the training set?", "_____no_output_____" ] ], [ [ "# Use this cell to examine one of the processed reviews to make sure everything is working as intended.\nprint(len(train_X))\nprint(len(train_X_len))\nprint(len(test_X))\nprint(len(test_X_len))\nprint(len(train_X[100]))\nprint(train_X[100])", "25000\n25000\n25000\n25000\n500\n[ 421 1135 141 12 4 169 866 1219 1180 119 71 114 248 262\n 51 16 227 84 570 729 1067 3953 1 1 62 569 267 945\n 747 225 1020 629 3519 2255 386 4562 1695 506 4407 702 144 4\n 402 900 296 32 131 196 92 8 151 154 2694 729 1161 5\n 50 12 322 729 181 86 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n 0 0 0 0 0 0 0 0 0 0]\n" ] ], [ [ "**Question:** In the cells above we use the `preprocess_data` and `convert_and_pad_data` methods to process both the training and testing set. Why or why not might this be a problem?", "_____no_output_____" ], [ "**Answer:** it can be a problem. we should only have access to the training set so our transformer can only use the training set to construct a representation. The picke file take long time to process until finished. preprocess_data faces problem when the pickle file is not ready yet.\nconvert_and_pad_data may have problem if the review is too long (now size ist limited to 500)-> we will lost the information. Maybe we extract the total different meaning of the original review, when the critical comments in the last part of the review.", "_____no_output_____" ], [ "## Step 3: Upload the data to S3\n\nAs in the XGBoost notebook, we will need to upload the training dataset to S3 in order for our training code to access it. For now we will save it locally and we will upload to S3 later on.\n\n### Save the processed training dataset locally\n\nIt is important to note the format of the data that we are saving as we will need to know it when we write the training code. In our case, each row of the dataset has the form `label`, `length`, `review[500]` where `review[500]` is a sequence of `500` integers representing the words in the review.", "_____no_output_____" ] ], [ [ "import pandas as pd\n \npd.concat([pd.DataFrame(train_y), pd.DataFrame(train_X_len), pd.DataFrame(train_X)], axis=1) \\\n .to_csv(os.path.join(data_dir, 'train.csv'), header=False, index=False)", "_____no_output_____" ] ], [ [ "### Uploading the training data\n\n\nNext, we need to upload the training data to the SageMaker default S3 bucket so that we can provide access to it while training our model.", "_____no_output_____" ] ], [ [ "import sagemaker\n\nsagemaker_session = sagemaker.Session()\n\nbucket = sagemaker_session.default_bucket()\nprefix = 'sagemaker/sentiment_rnn'\n\nrole = sagemaker.get_execution_role()", "_____no_output_____" ], [ "input_data = sagemaker_session.upload_data(path=data_dir, bucket=bucket, key_prefix=prefix)", "_____no_output_____" ] ], [ [ "**NOTE:** The cell above uploads the entire contents of our data directory. This includes the `word_dict.pkl` file. This is fortunate as we will need this later on when we create an endpoint that accepts an arbitrary review. For now, we will just take note of the fact that it resides in the data directory (and so also in the S3 training bucket) and that we will need to make sure it gets saved in the model directory.", "_____no_output_____" ], [ "## Step 4: Build and Train the PyTorch Model\n\nIn the XGBoost notebook we discussed what a model is in the SageMaker framework. In particular, a model comprises three objects\n\n - Model Artifacts,\n - Training Code, and\n - Inference Code,\n \neach of which interact with one another. In the XGBoost example we used training and inference code that was provided by Amazon. Here we will still be using containers provided by Amazon with the added benefit of being able to include our own custom code.\n\nWe will start by implementing our own neural network in PyTorch along with a training script. For the purposes of this project we have provided the necessary model object in the `model.py` file, inside of the `train` folder. You can see the provided implementation by running the cell below.", "_____no_output_____" ] ], [ [ "!pygmentize train/model.py", "\u001b[34mimport\u001b[39;49;00m \u001b[04m\u001b[36mtorch\u001b[39;49;00m\u001b[04m\u001b[36m.\u001b[39;49;00m\u001b[04m\u001b[36mnn\u001b[39;49;00m \u001b[34mas\u001b[39;49;00m \u001b[04m\u001b[36mnn\u001b[39;49;00m\r\n\r\n\u001b[34mclass\u001b[39;49;00m \u001b[04m\u001b[32mLSTMClassifier\u001b[39;49;00m(nn.Module):\r\n \u001b[33m\"\"\"\u001b[39;49;00m\r\n\u001b[33m This is the simple RNN model we will be using to perform Sentiment Analysis.\u001b[39;49;00m\r\n\u001b[33m \"\"\"\u001b[39;49;00m\r\n\r\n \u001b[34mdef\u001b[39;49;00m \u001b[32m__init__\u001b[39;49;00m(\u001b[36mself\u001b[39;49;00m, embedding_dim, hidden_dim, vocab_size):\r\n \u001b[33m\"\"\"\u001b[39;49;00m\r\n\u001b[33m Initialize the model by settingg up the various layers.\u001b[39;49;00m\r\n\u001b[33m \"\"\"\u001b[39;49;00m\r\n \u001b[36msuper\u001b[39;49;00m(LSTMClassifier, \u001b[36mself\u001b[39;49;00m).\u001b[32m__init__\u001b[39;49;00m()\r\n\r\n \u001b[36mself\u001b[39;49;00m.embedding = nn.Embedding(vocab_size, embedding_dim, padding_idx=\u001b[34m0\u001b[39;49;00m)\r\n \u001b[36mself\u001b[39;49;00m.lstm = nn.LSTM(embedding_dim, hidden_dim)\r\n \u001b[36mself\u001b[39;49;00m.dense = nn.Linear(in_features=hidden_dim, out_features=\u001b[34m1\u001b[39;49;00m)\r\n \u001b[36mself\u001b[39;49;00m.sig = nn.Sigmoid()\r\n \r\n \u001b[36mself\u001b[39;49;00m.word_dict = \u001b[34mNone\u001b[39;49;00m\r\n\r\n \u001b[34mdef\u001b[39;49;00m \u001b[32mforward\u001b[39;49;00m(\u001b[36mself\u001b[39;49;00m, x):\r\n \u001b[33m\"\"\"\u001b[39;49;00m\r\n\u001b[33m Perform a forward pass of our model on some input.\u001b[39;49;00m\r\n\u001b[33m \"\"\"\u001b[39;49;00m\r\n x = x.t()\r\n lengths = x[\u001b[34m0\u001b[39;49;00m,:]\r\n reviews = x[\u001b[34m1\u001b[39;49;00m:,:]\r\n embeds = \u001b[36mself\u001b[39;49;00m.embedding(reviews)\r\n lstm_out, _ = \u001b[36mself\u001b[39;49;00m.lstm(embeds)\r\n out = \u001b[36mself\u001b[39;49;00m.dense(lstm_out)\r\n out = out[lengths - \u001b[34m1\u001b[39;49;00m, \u001b[36mrange\u001b[39;49;00m(\u001b[36mlen\u001b[39;49;00m(lengths))]\r\n \u001b[34mreturn\u001b[39;49;00m \u001b[36mself\u001b[39;49;00m.sig(out.squeeze())\r\n" ] ], [ [ "The important takeaway from the implementation provided is that there are three parameters that we may wish to tweak to improve the performance of our model. These are the embedding dimension, the hidden dimension and the size of the vocabulary. We will likely want to make these parameters configurable in the training script so that if we wish to modify them we do not need to modify the script itself. We will see how to do this later on. To start we will write some of the training code in the notebook so that we can more easily diagnose any issues that arise.\n\nFirst we will load a small portion of the training data set to use as a sample. It would be very time consuming to try and train the model completely in the notebook as we do not have access to a gpu and the compute instance that we are using is not particularly powerful. However, we can work on a small bit of the data to get a feel for how our training script is behaving.", "_____no_output_____" ] ], [ [ "import torch\nimport torch.utils.data\n\n# Read in only the first 250 rows\ntrain_sample = pd.read_csv(os.path.join(data_dir, 'train.csv'), header=None, names=None, nrows=250)\n\n# Turn the input pandas dataframe into tensors\ntrain_sample_y = torch.from_numpy(train_sample[[0]].values).float().squeeze()\ntrain_sample_X = torch.from_numpy(train_sample.drop([0], axis=1).values).long()\n\n# Build the dataset\ntrain_sample_ds = torch.utils.data.TensorDataset(train_sample_X, train_sample_y)\n# Build the dataloader\ntrain_sample_dl = torch.utils.data.DataLoader(train_sample_ds, batch_size=50)", "_____no_output_____" ] ], [ [ "### (TODO) Writing the training method\n\nNext we need to write the training code itself. This should be very similar to training methods that you have written before to train PyTorch models. We will leave any difficult aspects such as model saving / loading and parameter loading until a little later.", "_____no_output_____" ] ], [ [ "def train(model, train_loader, epochs, optimizer, loss_fn, device):\n for epoch in range(1, epochs + 1):\n model.train()\n total_loss = 0\n for batch in train_loader: \n batch_X, batch_y = batch\n \n batch_X = batch_X.to(device)\n batch_y = batch_y.to(device)\n \n # TODO: Complete this train method to train the model provided.\n optimizer.zero_grad()\n out = model(batch_X)\n #model.zero_grad()\n #out = model.forward(batch_X)\n loss = loss_fn(out, batch_y)\n loss.backward()\n optimizer.step()\n total_loss += loss.data.item()\n print(\"Epoch: {}, BCELoss: {}\".format(epoch, total_loss / len(train_loader)))", "_____no_output_____" ] ], [ [ "Supposing we have the training method above, we will test that it is working by writing a bit of code in the notebook that executes our training method on the small sample training set that we loaded earlier. The reason for doing this in the notebook is so that we have an opportunity to fix any errors that arise early when they are easier to diagnose.", "_____no_output_____" ] ], [ [ "import torch.optim as optim\nfrom train.model import LSTMClassifier\n\ndevice = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\nmodel = LSTMClassifier(32, 100, 5000).to(device)\noptimizer = optim.Adam(model.parameters())\nloss_fn = torch.nn.BCELoss()\n\ntrain(model, train_sample_dl, 5, optimizer, loss_fn, device)", "Epoch: 1, BCELoss: 0.6912282586097718\nEpoch: 2, BCELoss: 0.6808721899986268\nEpoch: 3, BCELoss: 0.6719786524772644\nEpoch: 4, BCELoss: 0.6623488545417786\nEpoch: 5, BCELoss: 0.65109281539917\n" ] ], [ [ "In order to construct a PyTorch model using SageMaker we must provide SageMaker with a training script. We may optionally include a directory which will be copied to the container and from which our training code will be run. When the training container is executed it will check the uploaded directory (if there is one) for a `requirements.txt` file and install any required Python libraries, after which the training script will be run.", "_____no_output_____" ], [ "### (TODO) Training the model\n\nWhen a PyTorch model is constructed in SageMaker, an entry point must be specified. This is the Python file which will be executed when the model is trained. Inside of the `train` directory is a file called `train.py` which has been provided and which contains most of the necessary code to train our model. The only thing that is missing is the implementation of the `train()` method which you wrote earlier in this notebook.\n\n**TODO**: Copy the `train()` method written above and paste it into the `train/train.py` file where required.\n\nThe way that SageMaker passes hyperparameters to the training script is by way of arguments. These arguments can then be parsed and used in the training script. To see how this is done take a look at the provided `train/train.py` file.", "_____no_output_____" ] ], [ [ "from sagemaker.pytorch import PyTorch\n\nestimator = PyTorch(entry_point=\"train.py\",\n source_dir=\"train\",\n role=role,\n framework_version='0.4.0',\n train_instance_count=1,\n train_instance_type='ml.p2.xlarge', #original: train_instance_type='ml.p2.xlarge', the support center haven't processed my ticket\n hyperparameters={\n 'epochs': 10,\n 'hidden_dim': 200,\n })", "_____no_output_____" ], [ "estimator.fit({'training': input_data})", "'create_image_uri' will be deprecated in favor of 'ImageURIProvider' class in SageMaker Python SDK v2.\n's3_input' class will be renamed to 'TrainingInput' in SageMaker Python SDK v2.\n'create_image_uri' will be deprecated in favor of 'ImageURIProvider' class in SageMaker Python SDK v2.\n" ] ], [ [ "## Step 5: Testing the model\n\nAs mentioned at the top of this notebook, we will be testing this model by first deploying it and then sending the testing data to the deployed endpoint. We will do this so that we can make sure that the deployed model is working correctly.\n\n## Step 6: Deploy the model for testing\n\nNow that we have trained our model, we would like to test it to see how it performs. Currently our model takes input of the form `review_length, review[500]` where `review[500]` is a sequence of `500` integers which describe the words present in the review, encoded using `word_dict`. Fortunately for us, SageMaker provides built-in inference code for models with simple inputs such as this.\n\nThere is one thing that we need to provide, however, and that is a function which loads the saved model. This function must be called `model_fn()` and takes as its only parameter a path to the directory where the model artifacts are stored. This function must also be present in the python file which we specified as the entry point. In our case the model loading function has been provided and so no changes need to be made.\n\n**NOTE**: When the built-in inference code is run it must import the `model_fn()` method from the `train.py` file. This is why the training code is wrapped in a main guard ( ie, `if __name__ == '__main__':` )\n\nSince we don't need to change anything in the code that was uploaded during training, we can simply deploy the current model as-is.\n\n**NOTE:** When deploying a model you are asking SageMaker to launch an compute instance that will wait for data to be sent to it. As a result, this compute instance will continue to run until *you* shut it down. This is important to know since the cost of a deployed endpoint depends on how long it has been running for.\n\nIn other words **If you are no longer using a deployed endpoint, shut it down!**\n\n**TODO:** Deploy the trained model.", "_____no_output_____" ] ], [ [ "estimator.model_data", "_____no_output_____" ], [ "training_job_name=estimator.latest_training_job.name", "_____no_output_____" ], [ "# TODO: Deploy the trained model\npredictor = estimator.deploy(initial_instance_count = 1, instance_type = 'ml.m4.xlarge')\n#original: train_instance_type='ml.p2.xlarge', the support center haven't processed my ticket", "Parameter image will be renamed to image_uri in SageMaker Python SDK v2.\n'create_image_uri' will be deprecated in favor of 'ImageURIProvider' class in SageMaker Python SDK v2.\nUsing already existing model: sagemaker-pytorch-2020-10-08-10-19-04-219\n" ], [ "model_predict = PyTorchModel(model_data=estimator.model_data,\n role = role,\n framework_version='0.4.0',\n entry_point=\"train.py\",\n source_dir=\"train\")\npredictor_predict = model_predict.deploy(initial_instance_count=1, instance_type='ml.m4.xlarge')", "Parameter image will be renamed to image_uri in SageMaker Python SDK v2.\n'create_image_uri' will be deprecated in favor of 'ImageURIProvider' class in SageMaker Python SDK v2.\n" ], [ "predictor_predict.endpoint", "_____no_output_____" ], [ "predictor = estimator.deploy(initial_instance_count = 1, instance_type = 'ml.p2.xlarge')", "Parameter image will be renamed to image_uri in SageMaker Python SDK v2.\n'create_image_uri' will be deprecated in favor of 'ImageURIProvider' class in SageMaker Python SDK v2.\n" ] ], [ [ "## Step 7 - Use the model for testing\n\nOnce deployed, we can read in the test data and send it off to our deployed model to get some results. Once we collect all of the results we can determine how accurate our model is.", "_____no_output_____" ] ], [ [ "test_X = pd.concat([pd.DataFrame(test_X_len), pd.DataFrame(test_X)], axis=1)", "_____no_output_____" ], [ "# We split the data into chunks and send each chunk seperately, accumulating the results.\n\ndef predict(data, rows=512):\n split_array = np.array_split(data, int(data.shape[0] / float(rows) + 1))\n predictions = np.array([])\n for array in split_array:\n predictions = np.append(predictions, predictor.predict(array))\n \n return predictions", "_____no_output_____" ], [ "predictions = predict(test_X.values)\npredictions = [round(num) for num in predictions]", "_____no_output_____" ], [ "from sklearn.metrics import accuracy_score\naccuracy_score(test_y, predictions)", "_____no_output_____" ] ], [ [ "**Question:** How does this model compare to the XGBoost model you created earlier? Why might these two models perform differently on this dataset? Which do *you* think is better for sentiment analysis?", "_____no_output_____" ], [ "**Answer:**XGBoost has a better accuray score than this model and run much faster. The trainning job for XGBoost takes only 3 minutes, but this model takes 2 hours. I think XGBoost is better for sentiment analysis. Actually this two accuray score is close to each other, may this model can be improved somehow.", "_____no_output_____" ], [ "### (TODO) More testing\n\nWe now have a trained model which has been deployed and which we can send processed reviews to and which returns the predicted sentiment. However, ultimately we would like to be able to send our model an unprocessed review. That is, we would like to send the review itself as a string. For example, suppose we wish to send the following review to our model.", "_____no_output_____" ] ], [ [ "test_review = 'The simplest pleasures in life are the best, and this film is one of them. Combining a rather basic storyline of love and adventure this movie transcends the usual weekend fair with wit and unmitigated charm.'", "_____no_output_____" ] ], [ [ "The question we now need to answer is, how do we send this review to our model?\n\nRecall in the first section of this notebook we did a bunch of data processing to the IMDb dataset. In particular, we did two specific things to the provided reviews.\n - Removed any html tags and stemmed the input\n - Encoded the review as a sequence of integers using `word_dict`\n \nIn order process the review we will need to repeat these two steps.\n\n**TODO**: Using the `review_to_words` and `convert_and_pad` methods from section one, convert `test_review` into a numpy array `test_data` suitable to send to our model. Remember that our model expects input of the form `review_length, review[500]`.", "_____no_output_____" ] ], [ [ "# TODO: Convert test_review into a form usable by the model and save the results in test_data\ntest_data = None\ntest_words = review_to_words(test_review)\ntest_data, test_data_len = convert_and_pad(word_dict, test_words)", "_____no_output_____" ] ], [ [ "Now that we have processed the review, we can send the resulting array to our model to predict the sentiment of the review.", "_____no_output_____" ] ], [ [ "predictor.predict(test_data)", "_____no_output_____" ] ], [ [ "Since the return value of our model is close to `1`, we can be certain that the review we submitted is positive.", "_____no_output_____" ], [ "### Delete the endpoint\n\nOf course, just like in the XGBoost notebook, once we've deployed an endpoint it continues to run until we tell it to shut down. Since we are done using our endpoint for now, we can delete it.", "_____no_output_____" ] ], [ [ "estimator.delete_endpoint()", "_____no_output_____" ] ], [ [ "## Step 6 (again) - Deploy the model for the web app\n\nNow that we know that our model is working, it's time to create some custom inference code so that we can send the model a review which has not been processed and have it determine the sentiment of the review.\n\nAs we saw above, by default the estimator which we created, when deployed, will use the entry script and directory which we provided when creating the model. However, since we now wish to accept a string as input and our model expects a processed review, we need to write some custom inference code.\n\nWe will store the code that we write in the `serve` directory. Provided in this directory is the `model.py` file that we used to construct our model, a `utils.py` file which contains the `review_to_words` and `convert_and_pad` pre-processing functions which we used during the initial data processing, and `predict.py`, the file which will contain our custom inference code. Note also that `requirements.txt` is present which will tell SageMaker what Python libraries are required by our custom inference code.\n\nWhen deploying a PyTorch model in SageMaker, you are expected to provide four functions which the SageMaker inference container will use.\n - `model_fn`: This function is the same function that we used in the training script and it tells SageMaker how to load our model.\n - `input_fn`: This function receives the raw serialized input that has been sent to the model's endpoint and its job is to de-serialize and make the input available for the inference code.\n - `output_fn`: This function takes the output of the inference code and its job is to serialize this output and return it to the caller of the model's endpoint.\n - `predict_fn`: The heart of the inference script, this is where the actual prediction is done and is the function which you will need to complete.\n\nFor the simple website that we are constructing during this project, the `input_fn` and `output_fn` methods are relatively straightforward. We only require being able to accept a string as input and we expect to return a single value as output. You might imagine though that in a more complex application the input or output may be image data or some other binary data which would require some effort to serialize.\n\n### (TODO) Writing inference code\n\nBefore writing our custom inference code, we will begin by taking a look at the code which has been provided.", "_____no_output_____" ] ], [ [ "!pygmentize serve/predict.py", "\u001b[34mimport\u001b[39;49;00m \u001b[04m\u001b[36margparse\u001b[39;49;00m\r\n\u001b[34mimport\u001b[39;49;00m \u001b[04m\u001b[36mjson\u001b[39;49;00m\r\n\u001b[34mimport\u001b[39;49;00m \u001b[04m\u001b[36mos\u001b[39;49;00m\r\n\u001b[34mimport\u001b[39;49;00m \u001b[04m\u001b[36mpickle\u001b[39;49;00m\r\n\u001b[34mimport\u001b[39;49;00m \u001b[04m\u001b[36msys\u001b[39;49;00m\r\n\u001b[34mimport\u001b[39;49;00m \u001b[04m\u001b[36msagemaker_containers\u001b[39;49;00m\r\n\u001b[34mimport\u001b[39;49;00m \u001b[04m\u001b[36mpandas\u001b[39;49;00m \u001b[34mas\u001b[39;49;00m \u001b[04m\u001b[36mpd\u001b[39;49;00m\r\n\u001b[34mimport\u001b[39;49;00m \u001b[04m\u001b[36mnumpy\u001b[39;49;00m \u001b[34mas\u001b[39;49;00m \u001b[04m\u001b[36mnp\u001b[39;49;00m\r\n\u001b[34mimport\u001b[39;49;00m \u001b[04m\u001b[36mtorch\u001b[39;49;00m\r\n\u001b[34mimport\u001b[39;49;00m \u001b[04m\u001b[36mtorch\u001b[39;49;00m\u001b[04m\u001b[36m.\u001b[39;49;00m\u001b[04m\u001b[36mnn\u001b[39;49;00m \u001b[34mas\u001b[39;49;00m \u001b[04m\u001b[36mnn\u001b[39;49;00m\r\n\u001b[34mimport\u001b[39;49;00m \u001b[04m\u001b[36mtorch\u001b[39;49;00m\u001b[04m\u001b[36m.\u001b[39;49;00m\u001b[04m\u001b[36moptim\u001b[39;49;00m \u001b[34mas\u001b[39;49;00m \u001b[04m\u001b[36moptim\u001b[39;49;00m\r\n\u001b[34mimport\u001b[39;49;00m \u001b[04m\u001b[36mtorch\u001b[39;49;00m\u001b[04m\u001b[36m.\u001b[39;49;00m\u001b[04m\u001b[36mutils\u001b[39;49;00m\u001b[04m\u001b[36m.\u001b[39;49;00m\u001b[04m\u001b[36mdata\u001b[39;49;00m\r\n\r\n\u001b[34mfrom\u001b[39;49;00m \u001b[04m\u001b[36mmodel\u001b[39;49;00m \u001b[34mimport\u001b[39;49;00m LSTMClassifier\r\n\r\n\u001b[34mfrom\u001b[39;49;00m \u001b[04m\u001b[36mutils\u001b[39;49;00m \u001b[34mimport\u001b[39;49;00m review_to_words, convert_and_pad\r\n\r\n\u001b[34mdef\u001b[39;49;00m \u001b[32mmodel_fn\u001b[39;49;00m(model_dir):\r\n \u001b[33m\"\"\"Load the PyTorch model from the `model_dir` directory.\"\"\"\u001b[39;49;00m\r\n \u001b[36mprint\u001b[39;49;00m(\u001b[33m\"\u001b[39;49;00m\u001b[33mLoading model.\u001b[39;49;00m\u001b[33m\"\u001b[39;49;00m)\r\n\r\n \u001b[37m# First, load the parameters used to create the model.\u001b[39;49;00m\r\n model_info = {}\r\n model_info_path = os.path.join(model_dir, \u001b[33m'\u001b[39;49;00m\u001b[33mmodel_info.pth\u001b[39;49;00m\u001b[33m'\u001b[39;49;00m)\r\n \u001b[34mwith\u001b[39;49;00m \u001b[36mopen\u001b[39;49;00m(model_info_path, \u001b[33m'\u001b[39;49;00m\u001b[33mrb\u001b[39;49;00m\u001b[33m'\u001b[39;49;00m) \u001b[34mas\u001b[39;49;00m f:\r\n model_info = torch.load(f)\r\n\r\n \u001b[36mprint\u001b[39;49;00m(\u001b[33m\"\u001b[39;49;00m\u001b[33mmodel_info: \u001b[39;49;00m\u001b[33m{}\u001b[39;49;00m\u001b[33m\"\u001b[39;49;00m.format(model_info))\r\n\r\n \u001b[37m# Determine the device and construct the model.\u001b[39;49;00m\r\n device = torch.device(\u001b[33m\"\u001b[39;49;00m\u001b[33mcuda\u001b[39;49;00m\u001b[33m\"\u001b[39;49;00m \u001b[34mif\u001b[39;49;00m torch.cuda.is_available() \u001b[34melse\u001b[39;49;00m \u001b[33m\"\u001b[39;49;00m\u001b[33mcpu\u001b[39;49;00m\u001b[33m\"\u001b[39;49;00m)\r\n model = LSTMClassifier(model_info[\u001b[33m'\u001b[39;49;00m\u001b[33membedding_dim\u001b[39;49;00m\u001b[33m'\u001b[39;49;00m], model_info[\u001b[33m'\u001b[39;49;00m\u001b[33mhidden_dim\u001b[39;49;00m\u001b[33m'\u001b[39;49;00m], model_info[\u001b[33m'\u001b[39;49;00m\u001b[33mvocab_size\u001b[39;49;00m\u001b[33m'\u001b[39;49;00m])\r\n\r\n \u001b[37m# Load the store model parameters.\u001b[39;49;00m\r\n model_path = os.path.join(model_dir, \u001b[33m'\u001b[39;49;00m\u001b[33mmodel.pth\u001b[39;49;00m\u001b[33m'\u001b[39;49;00m)\r\n \u001b[34mwith\u001b[39;49;00m \u001b[36mopen\u001b[39;49;00m(model_path, \u001b[33m'\u001b[39;49;00m\u001b[33mrb\u001b[39;49;00m\u001b[33m'\u001b[39;49;00m) \u001b[34mas\u001b[39;49;00m f:\r\n model.load_state_dict(torch.load(f))\r\n\r\n \u001b[37m# Load the saved word_dict.\u001b[39;49;00m\r\n word_dict_path = os.path.join(model_dir, \u001b[33m'\u001b[39;49;00m\u001b[33mword_dict.pkl\u001b[39;49;00m\u001b[33m'\u001b[39;49;00m)\r\n \u001b[34mwith\u001b[39;49;00m \u001b[36mopen\u001b[39;49;00m(word_dict_path, \u001b[33m'\u001b[39;49;00m\u001b[33mrb\u001b[39;49;00m\u001b[33m'\u001b[39;49;00m) \u001b[34mas\u001b[39;49;00m f:\r\n model.word_dict = pickle.load(f)\r\n\r\n model.to(device).eval()\r\n\r\n \u001b[36mprint\u001b[39;49;00m(\u001b[33m\"\u001b[39;49;00m\u001b[33mDone loading model.\u001b[39;49;00m\u001b[33m\"\u001b[39;49;00m)\r\n \u001b[34mreturn\u001b[39;49;00m model\r\n\r\n\u001b[34mdef\u001b[39;49;00m \u001b[32minput_fn\u001b[39;49;00m(serialized_input_data, content_type):\r\n \u001b[36mprint\u001b[39;49;00m(\u001b[33m'\u001b[39;49;00m\u001b[33mDeserializing the input data.\u001b[39;49;00m\u001b[33m'\u001b[39;49;00m)\r\n \u001b[34mif\u001b[39;49;00m content_type == \u001b[33m'\u001b[39;49;00m\u001b[33mtext/plain\u001b[39;49;00m\u001b[33m'\u001b[39;49;00m:\r\n data = serialized_input_data.decode(\u001b[33m'\u001b[39;49;00m\u001b[33mutf-8\u001b[39;49;00m\u001b[33m'\u001b[39;49;00m)\r\n \u001b[34mreturn\u001b[39;49;00m data\r\n \u001b[34mraise\u001b[39;49;00m \u001b[36mException\u001b[39;49;00m(\u001b[33m'\u001b[39;49;00m\u001b[33mRequested unsupported ContentType in content_type: \u001b[39;49;00m\u001b[33m'\u001b[39;49;00m + content_type)\r\n\r\n\u001b[34mdef\u001b[39;49;00m \u001b[32moutput_fn\u001b[39;49;00m(prediction_output, accept):\r\n \u001b[36mprint\u001b[39;49;00m(\u001b[33m'\u001b[39;49;00m\u001b[33mSerializing the generated output.\u001b[39;49;00m\u001b[33m'\u001b[39;49;00m)\r\n \u001b[34mreturn\u001b[39;49;00m \u001b[36mstr\u001b[39;49;00m(prediction_output)\r\n\r\n\u001b[34mdef\u001b[39;49;00m \u001b[32mpredict_fn\u001b[39;49;00m(input_data, model):\r\n \u001b[36mprint\u001b[39;49;00m(\u001b[33m'\u001b[39;49;00m\u001b[33mInferring sentiment of input data.\u001b[39;49;00m\u001b[33m'\u001b[39;49;00m)\r\n\r\n device = torch.device(\u001b[33m\"\u001b[39;49;00m\u001b[33mcuda\u001b[39;49;00m\u001b[33m\"\u001b[39;49;00m \u001b[34mif\u001b[39;49;00m torch.cuda.is_available() \u001b[34melse\u001b[39;49;00m \u001b[33m\"\u001b[39;49;00m\u001b[33mcpu\u001b[39;49;00m\u001b[33m\"\u001b[39;49;00m)\r\n \r\n \u001b[34mif\u001b[39;49;00m model.word_dict \u001b[35mis\u001b[39;49;00m \u001b[34mNone\u001b[39;49;00m:\r\n \u001b[34mraise\u001b[39;49;00m \u001b[36mException\u001b[39;49;00m(\u001b[33m'\u001b[39;49;00m\u001b[33mModel has not been loaded properly, no word_dict.\u001b[39;49;00m\u001b[33m'\u001b[39;49;00m)\r\n \r\n \u001b[37m# TODO: Process input_data so that it is ready to be sent to our model.\u001b[39;49;00m\r\n \u001b[37m# You should produce two variables:\u001b[39;49;00m\r\n \u001b[37m# data_X - A sequence of length 500 which represents the converted review\u001b[39;49;00m\r\n \u001b[37m# data_len - The length of the review\u001b[39;49;00m\r\n \r\n data_X = \u001b[34mNone\u001b[39;49;00m\r\n data_len = \u001b[34mNone\u001b[39;49;00m\r\n words = review_to_words(input_data)\r\n data_X, data_len = convert_and_pad(model.word_dict, words)\r\n \u001b[37m# Using data_X and data_len we construct an appropriate input tensor. Remember\u001b[39;49;00m\r\n \u001b[37m# that our model expects input data of the form 'len, review[500]'.\u001b[39;49;00m\r\n data_pack = np.hstack((data_len, data_X))\r\n data_pack = data_pack.reshape(\u001b[34m1\u001b[39;49;00m, -\u001b[34m1\u001b[39;49;00m)\r\n \r\n data = torch.from_numpy(data_pack)\r\n data = data.to(device)\r\n\r\n \u001b[37m# Make sure to put the model into evaluation mode\u001b[39;49;00m\r\n model.eval()\r\n\r\n \u001b[37m# TODO: Compute the result of applying the model to the input data. The variable `result` should\u001b[39;49;00m\r\n \u001b[37m# be a numpy array which contains a single integer which is either 1 or 0\u001b[39;49;00m\r\n\r\n result = \u001b[34mNone\u001b[39;49;00m\r\n \u001b[34mwith\u001b[39;49;00m torch.no_grad():\r\n output = model.forward(data)\r\n result = \u001b[36mint\u001b[39;49;00m(np.round(output.numpy()))\r\n \u001b[34mreturn\u001b[39;49;00m result\r\n" ] ], [ [ "As mentioned earlier, the `model_fn` method is the same as the one provided in the training code and the `input_fn` and `output_fn` methods are very simple and your task will be to complete the `predict_fn` method. Make sure that you save the completed file as `predict.py` in the `serve` directory.\n\n**TODO**: Complete the `predict_fn()` method in the `serve/predict.py` file.", "_____no_output_____" ], [ "### Deploying the model\n\nNow that the custom inference code has been written, we will create and deploy our model. To begin with, we need to construct a new PyTorchModel object which points to the model artifacts created during training and also points to the inference code that we wish to use. Then we can call the deploy method to launch the deployment container.\n\n**NOTE**: The default behaviour for a deployed PyTorch model is to assume that any input passed to the predictor is a `numpy` array. In our case we want to send a string so we need to construct a simple wrapper around the `RealTimePredictor` class to accomodate simple strings. In a more complicated situation you may want to provide a serialization object, for example if you wanted to sent image data.", "_____no_output_____" ] ], [ [ "from sagemaker.predictor import RealTimePredictor\nfrom sagemaker.pytorch import PyTorchModel\n\nclass StringPredictor(RealTimePredictor):\n def __init__(self, endpoint_name, sagemaker_session):\n super(StringPredictor, self).__init__(endpoint_name, sagemaker_session, content_type='text/plain')\n\nmodel = PyTorchModel(model_data=estimator.model_data,\n role = role,\n framework_version='0.4.0',\n entry_point='predict.py',\n source_dir='serve',\n predictor_cls=StringPredictor)\npredictor = model.deploy(initial_instance_count=1, instance_type='ml.m4.xlarge')", "Parameter image will be renamed to image_uri in SageMaker Python SDK v2.\n'create_image_uri' will be deprecated in favor of 'ImageURIProvider' class in SageMaker Python SDK v2.\n" ] ], [ [ "### Testing the model\n\nNow that we have deployed our model with the custom inference code, we should test to see if everything is working. Here we test our model by loading the first `250` positive and negative reviews and send them to the endpoint, then collect the results. The reason for only sending some of the data is that the amount of time it takes for our model to process the input and then perform inference is quite long and so testing the entire data set would be prohibitive.", "_____no_output_____" ] ], [ [ "import glob\n\ndef test_reviews(data_dir='../data/aclImdb', stop=250):\n \n results = []\n ground = []\n \n # We make sure to test both positive and negative reviews \n for sentiment in ['pos', 'neg']:\n \n path = os.path.join(data_dir, 'test', sentiment, '*.txt')\n files = glob.glob(path)\n \n files_read = 0\n \n print('Starting ', sentiment, ' files')\n \n # Iterate through the files and send them to the predictor\n for f in files:\n with open(f) as review:\n # First, we store the ground truth (was the review positive or negative)\n if sentiment == 'pos':\n ground.append(1)\n else:\n ground.append(0)\n # Read in the review and convert to 'utf-8' for transmission via HTTP\n review_input = review.read().encode('utf-8')\n # Send the review to the predictor and store the results\n results.append(int(predictor.predict(review_input)))\n \n # Sending reviews to our endpoint one at a time takes a while so we\n # only send a small number of reviews\n files_read += 1\n if files_read == stop:\n break\n \n return ground, results", "_____no_output_____" ], [ "ground, results = test_reviews()", "Starting pos files\nStarting neg files\n" ], [ "from sklearn.metrics import accuracy_score\naccuracy_score(ground, results)", "_____no_output_____" ] ], [ [ "As an additional test, we can try sending the `test_review` that we looked at earlier.", "_____no_output_____" ] ], [ [ "predictor.predict(test_review)", "_____no_output_____" ] ], [ [ "Now that we know our endpoint is working as expected, we can set up the web page that will interact with it. If you don't have time to finish the project now, make sure to skip down to the end of this notebook and shut down your endpoint. You can deploy it again when you come back.", "_____no_output_____" ], [ "## Step 7 (again): Use the model for the web app\n\n> **TODO:** This entire section and the next contain tasks for you to complete, mostly using the AWS console.\n\nSo far we have been accessing our model endpoint by constructing a predictor object which uses the endpoint and then just using the predictor object to perform inference. What if we wanted to create a web app which accessed our model? The way things are set up currently makes that not possible since in order to access a SageMaker endpoint the app would first have to authenticate with AWS using an IAM role which included access to SageMaker endpoints. However, there is an easier way! We just need to use some additional AWS services.\n\n<img src=\"Web App Diagram.svg\">\n\nThe diagram above gives an overview of how the various services will work together. On the far right is the model which we trained above and which is deployed using SageMaker. On the far left is our web app that collects a user's movie review, sends it off and expects a positive or negative sentiment in return.\n\nIn the middle is where some of the magic happens. We will construct a Lambda function, which you can think of as a straightforward Python function that can be executed whenever a specified event occurs. We will give this function permission to send and recieve data from a SageMaker endpoint.\n\nLastly, the method we will use to execute the Lambda function is a new endpoint that we will create using API Gateway. This endpoint will be a url that listens for data to be sent to it. Once it gets some data it will pass that data on to the Lambda function and then return whatever the Lambda function returns. Essentially it will act as an interface that lets our web app communicate with the Lambda function.\n\n### Setting up a Lambda function\n\nThe first thing we are going to do is set up a Lambda function. This Lambda function will be executed whenever our public API has data sent to it. When it is executed it will receive the data, perform any sort of processing that is required, send the data (the review) to the SageMaker endpoint we've created and then return the result.\n\n#### Part A: Create an IAM Role for the Lambda function\n\nSince we want the Lambda function to call a SageMaker endpoint, we need to make sure that it has permission to do so. To do this, we will construct a role that we can later give the Lambda function.\n\nUsing the AWS Console, navigate to the **IAM** page and click on **Roles**. Then, click on **Create role**. Make sure that the **AWS service** is the type of trusted entity selected and choose **Lambda** as the service that will use this role, then click **Next: Permissions**.\n\nIn the search box type `sagemaker` and select the check box next to the **AmazonSageMakerFullAccess** policy. Then, click on **Next: Review**.\n\nLastly, give this role a name. Make sure you use a name that you will remember later on, for example `LambdaSageMakerRole`. Then, click on **Create role**.\n\n#### Part B: Create a Lambda function\n\nNow it is time to actually create the Lambda function.\n\nUsing the AWS Console, navigate to the AWS Lambda page and click on **Create a function**. When you get to the next page, make sure that **Author from scratch** is selected. Now, name your Lambda function, using a name that you will remember later on, for example `sentiment_analysis_func`. Make sure that the **Python 3.6** runtime is selected and then choose the role that you created in the previous part. Then, click on **Create Function**.\n\nOn the next page you will see some information about the Lambda function you've just created. If you scroll down you should see an editor in which you can write the code that will be executed when your Lambda function is triggered. In our example, we will use the code below. \n\n```python\n# We need to use the low-level library to interact with SageMaker since the SageMaker API\n# is not available natively through Lambda.\nimport boto3\n\ndef lambda_handler(event, context):\n\n # The SageMaker runtime is what allows us to invoke the endpoint that we've created.\n runtime = boto3.Session().client('sagemaker-runtime')\n\n # Now we use the SageMaker runtime to invoke our endpoint, sending the review we were given\n response = runtime.invoke_endpoint(EndpointName = '**ENDPOINT NAME HERE**', # The name of the endpoint we created\n ContentType = 'text/plain', # The data format that is expected\n Body = event['body']) # The actual review\n\n # The response is an HTTP response whose body contains the result of our inference\n result = response['Body'].read().decode('utf-8')\n\n return {\n 'statusCode' : 200,\n 'headers' : { 'Content-Type' : 'text/plain', 'Access-Control-Allow-Origin' : '*' },\n 'body' : result\n }\n```\n\nOnce you have copy and pasted the code above into the Lambda code editor, replace the `**ENDPOINT NAME HERE**` portion with the name of the endpoint that we deployed earlier. You can determine the name of the endpoint using the code cell below.", "_____no_output_____" ] ], [ [ "predictor.endpoint", "_____no_output_____" ] ], [ [ "Once you have added the endpoint name to the Lambda function, click on **Save**. Your Lambda function is now up and running. Next we need to create a way for our web app to execute the Lambda function.\n\n### Setting up API Gateway\n\nNow that our Lambda function is set up, it is time to create a new API using API Gateway that will trigger the Lambda function we have just created.\n\nUsing AWS Console, navigate to **Amazon API Gateway** and then click on **Get started**.\n\nOn the next page, make sure that **New API** is selected and give the new api a name, for example, `sentiment_analysis_api`. Then, click on **Create API**.\n\nNow we have created an API, however it doesn't currently do anything. What we want it to do is to trigger the Lambda function that we created earlier.\n\nSelect the **Actions** dropdown menu and click **Create Method**. A new blank method will be created, select its dropdown menu and select **POST**, then click on the check mark beside it.\n\nFor the integration point, make sure that **Lambda Function** is selected and click on the **Use Lambda Proxy integration**. This option makes sure that the data that is sent to the API is then sent directly to the Lambda function with no processing. It also means that the return value must be a proper response object as it will also not be processed by API Gateway.\n\nType the name of the Lambda function you created earlier into the **Lambda Function** text entry box and then click on **Save**. Click on **OK** in the pop-up box that then appears, giving permission to API Gateway to invoke the Lambda function you created.\n\nThe last step in creating the API Gateway is to select the **Actions** dropdown and click on **Deploy API**. You will need to create a new Deployment stage and name it anything you like, for example `prod`.\n\nYou have now successfully set up a public API to access your SageMaker model. Make sure to copy or write down the URL provided to invoke your newly created public API as this will be needed in the next step. This URL can be found at the top of the page, highlighted in blue next to the text **Invoke URL**.", "_____no_output_____" ], [ "## Step 4: Deploying our web app\n\nNow that we have a publicly available API, we can start using it in a web app. For our purposes, we have provided a simple static html file which can make use of the public api you created earlier.\n\nIn the `website` folder there should be a file called `index.html`. Download the file to your computer and open that file up in a text editor of your choice. There should be a line which contains **\\*\\*REPLACE WITH PUBLIC API URL\\*\\***. Replace this string with the url that you wrote down in the last step and then save the file.\n\nNow, if you open `index.html` on your local computer, your browser will behave as a local web server and you can use the provided site to interact with your SageMaker model.\n\nIf you'd like to go further, you can host this html file anywhere you'd like, for example using github or hosting a static site on Amazon's S3. Once you have done this you can share the link with anyone you'd like and have them play with it too!\n\n> **Important Note** In order for the web app to communicate with the SageMaker endpoint, the endpoint has to actually be deployed and running. This means that you are paying for it. Make sure that the endpoint is running when you want to use the web app but that you shut it down when you don't need it, otherwise you will end up with a surprisingly large AWS bill.\n\n**TODO:** Make sure that you include the edited `index.html` file in your project submission.", "_____no_output_____" ], [ "Now that your web app is working, trying playing around with it and see how well it works.\n\n**Question**: Give an example of a review that you entered into your web app. What was the predicted sentiment of your example review?", "_____no_output_____" ], [ "**Answer:**", "_____no_output_____" ], [ "### Delete the endpoint\n\nRemember to always shut down your endpoint if you are no longer using it. You are charged for the length of time that the endpoint is running so if you forget and leave it on you could end up with an unexpectedly large bill.", "_____no_output_____" ] ], [ [ "predictor.delete_endpoint()", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "## Statistics", "_____no_output_____" ], [ "### Questions", "_____no_output_____" ], [ "```{admonition} Problem: JOIN Dataframes\n:class: dropdown, tip\nCan you tell me the ways in which 2 pandas data frames can be joined?\n```", "_____no_output_____" ], [ "```{admonition} Solution:\n:class: dropdown\nA very high level difference is that merge() is used to combine two (or more) dataframes on the basis of values of common columns (indices can also be used, use left_index=True and/or right_index=True), and concat() is used to append one (or more) dataframes one below the other (or sideways, depending on whether the axis option is set to 0 or 1).\n\njoin() is used to merge 2 dataframes on the basis of the index; instead of using merge() with the option left_index=True we can use join().\n\n\n```", "_____no_output_____" ], [ "```{admonition} Problem: [GOOGLE] Normal Distribution\n:class: dropdown, tip\n\nWrite a function to generate N samples from a normal distribution and plot the histogram.\n```", "_____no_output_____" ] ], [ [ "import numpy as np\nimport matplotlib.pyplot as plt\nfrom scipy import stats\n\ndef normal_sample_generator(N):\n # can be done using np.random.randn or stats.norm.rvs\n #x = np.random.randn(N)\n x = stats.norm.rvs(size=N)\n num_bins = 20\n plt.hist(x, bins=num_bins, facecolor='blue', alpha=0.5)\n\n y = np.linspace(-4, 4, N)\n bin_width = (x.max() - x.min()) / num_bins\n plt.plot(y, stats.norm.pdf(y) * N * bin_width)\n\n plt.show()\n\nnormal_sample_generator(10000)", "_____no_output_____" ] ], [ [ "```{admonition} Problem: [UBER] Bernoulli trial generator\n:class: dropdown, tip\n\nGiven a random Bernoulli trial generator, write a function to return a value sampled from a normal distribution.\n```", "_____no_output_____" ], [ "```{admonition} Solution:\n:class: dropdown\nSolution pending, [Reference material link](Given a random Bernoulli trial generator, how do you return a value sampled from a normal distribution?)\n```", "_____no_output_____" ], [ "```{admonition} Problem: [PINTEREST] Interquartile Distance\n:class: dropdown, tip\n\nGiven an array of unsorted random numbers (decimals) find the interquartile distance.\n```", "_____no_output_____" ] ], [ [ "# Interquartile distance is the difference between first and third quartile\n\n# first let's generate a list of random numbers\n\nimport random\nimport numpy as np\n\nli = [round(random.uniform(33.33, 66.66), 2) for i in range(50)]\nprint(li)\n\nqtl_1 = np.quantile(li,.25)\nqtl_3 = np.quantile(li,.75)\n\nprint(\"Interquartile distance: \", qtl_1 - qtl_3)", "[54.81, 65.68, 63.85, 58.29, 60.14, 53.23, 52.58, 51.62, 61.6, 57.85, 51.37, 38.7, 35.87, 33.95, 61.65, 33.59, 61.33, 44.97, 62.49, 39.67, 51.03, 45.79, 60.99, 60.49, 64.8, 46.16, 46.61, 34.06, 37.78, 56.72, 39.62, 61.38, 55.27, 40.53, 49.31, 58.95, 37.49, 34.39, 60.47, 56.12, 61.41, 34.56, 58.18, 56.35, 63.59, 50.59, 61.51, 42.02, 52.43, 56.71]\nInterquartile distance: -17.7275\n" ] ], [ [ "````{admonition} Problem: [GENENTECH] Imputing the mdeian\n:class: dropdown, tip\n\nWrite a function cheese_median to impute the median price of the selected California cheeses in place of the missing values. You may assume at least one cheese is not missing its price.\n\nInput:\n\n```python\nimport pandas as pd\n\ncheeses = {\"Name\": [\"Bohemian Goat\", \"Central Coast Bleu\", \"Cowgirl Mozzarella\", \"Cypress Grove Cheddar\", \"Oakdale Colby\"], \"Price\" : [15.00, None, 30.00, None, 45.00]}\n\ndf_cheeses = pd.DataFrame(cheeses)\n```\n\n| Name | Price |\n|:---------------------:|:-----:|\n| Bohemian Goat | 15.00 |\n| Central Coast Bleu | 30.00 |\n| Cowgirl Mozzarella | 30.00 |\n| Cypress Grove Cheddar | 30.00 |\n| Oakdale Colby | 45.00 |\n\n````", "_____no_output_____" ] ], [ [ "import pandas as pd\n\ncheeses = {\"Name\": [\"Bohemian Goat\", \"Central Coast Bleu\", \"Cowgirl Mozzarella\", \"Cypress Grove Cheddar\", \"Oakdale Colby\"], \"Price\" : [15.00, None, 30.00, None, 45.00]}\n\ndf_cheeses = pd.DataFrame(cheeses)\n\ndf_cheeses['Price'] = df_cheeses['Price'].fillna(df_cheeses['Price'].median())\ndf_cheeses.head()", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Real Estate Price Prediction", "_____no_output_____" ] ], [ [ "import pandas as pd", "_____no_output_____" ], [ "df = pd.read_csv(\"data.csv\")", "_____no_output_____" ], [ "df.head()", "_____no_output_____" ], [ "df['CHAS'].value_counts()", "_____no_output_____" ], [ "df.info()", "<class 'pandas.core.frame.DataFrame'>\nRangeIndex: 506 entries, 0 to 505\nData columns (total 14 columns):\n # Column Non-Null Count Dtype \n--- ------ -------------- ----- \n 0 CRIM 506 non-null float64\n 1 ZN 506 non-null float64\n 2 INDUS 506 non-null float64\n 3 CHAS 506 non-null int64 \n 4 NOX 506 non-null float64\n 5 RM 506 non-null float64\n 6 AGE 506 non-null float64\n 7 DIS 506 non-null float64\n 8 RAD 506 non-null int64 \n 9 TAX 506 non-null int64 \n 10 PTRATIO 506 non-null float64\n 11 B 506 non-null float64\n 12 LSTAT 506 non-null float64\n 13 MEDV 506 non-null float64\ndtypes: float64(11), int64(3)\nmemory usage: 55.5 KB\n" ], [ "df.describe()", "_____no_output_____" ], [ "%matplotlib inline", "_____no_output_____" ], [ "import matplotlib.pyplot as plt", "_____no_output_____" ], [ "df.hist(bins=50, figsize=(20,15))", "_____no_output_____" ] ], [ [ "## train_test_split", "_____no_output_____" ] ], [ [ "import numpy as np\ndef split_train_test(data, test_ratio):\n np.random.seed(42)\n shuffled = np.random.permutation(len(data))\n test_set_size = int(len(data) * test_ratio)\n test_indices = shuffled[:test_set_size]\n train_indices = shuffled[test_set_size:]\n return data.iloc[train_indices], data.iloc[test_indices]", "_____no_output_____" ], [ "train_set, test_set = split_train_test(df, 0.2)", "_____no_output_____" ], [ "print(f\"The length of train dataset is: {len(train_set)}\")", "The length of train dataset is: 405\n" ], [ "print(f\"The length of train dataset is: {len(test_set)}\")", "The length of train dataset is: 101\n" ], [ "def data_percent_allocation(train_set, test_set):\n total = len(df)\n train_percent = round((len(train_set)/total) * 100)\n test_percent = round((len(test_set)/total) * 100)\n return train_percent, test_percent", "_____no_output_____" ], [ "data_percent_allocation(train_set, test_set)", "_____no_output_____" ] ], [ [ "## train_test_split from sklearn", "_____no_output_____" ] ], [ [ "from sklearn.model_selection import train_test_split\ntrain_set, test_set = train_test_split(df, test_size = 0.2, random_state = 42)\nprint(f\"The length of train dataset is: {len(train_set)}\")\nprint(f\"The length of train dataset is: {len(test_set)}\")", "The length of train dataset is: 404\nThe length of train dataset is: 102\n" ], [ "from sklearn.model_selection import StratifiedShuffleSplit\nsplit = StratifiedShuffleSplit(n_splits = 1, test_size = 0.2, random_state = 42)\nfor train_index, test_index in split.split(df, df['CHAS']):\n strat_train_set = df.loc[train_index]\n strat_test_set = df.loc[test_index] ", "_____no_output_____" ], [ "strat_test_set['CHAS'].value_counts()", "_____no_output_____" ], [ "test_set['CHAS'].value_counts()", "_____no_output_____" ], [ "strat_train_set['CHAS'].value_counts()", "_____no_output_____" ], [ "train_set['CHAS'].value_counts()", "_____no_output_____" ] ], [ [ "### Stratified learning equal splitting of zero and ones", "_____no_output_____" ] ], [ [ "95/7", "_____no_output_____" ], [ "376/28", "_____no_output_____" ], [ "df = strat_train_set.copy()", "_____no_output_____" ] ], [ [ "## Corelations", "_____no_output_____" ] ], [ [ "from pandas.plotting import scatter_matrix\nattributes = [\"MEDV\", \"RM\", \"ZN\" , \"LSTAT\"]\nscatter_matrix(df[attributes], figsize = (12,8))", "_____no_output_____" ], [ "df.plot(kind=\"scatter\", x=\"RM\", y=\"MEDV\", alpha=1)", "_____no_output_____" ] ], [ [ "### Trying out attribute combinations", "_____no_output_____" ] ], [ [ "df[\"TAXRM\"] = df[\"TAX\"]/df[\"RM\"]", "_____no_output_____" ], [ "df.head()", "_____no_output_____" ], [ "corr_matrix = df.corr()\ncorr_matrix['MEDV'].sort_values(ascending=False)\n# 1 means strong positive corr and -1 means strong negative corr.\n# EX: if RM will increase our final result(MEDV) in prediction will also increase.", "_____no_output_____" ], [ "df.plot(kind=\"scatter\", x=\"TAXRM\", y=\"MEDV\", alpha=1)", "_____no_output_____" ], [ "df = strat_train_set.drop(\"MEDV\", axis=1)\ndf_labels = strat_train_set[\"MEDV\"].copy()", "_____no_output_____" ] ], [ [ "## Pipeline", "_____no_output_____" ] ], [ [ "from sklearn.pipeline import Pipeline\nfrom sklearn.preprocessing import StandardScaler\nfrom sklearn.impute import SimpleImputer\n\nmy_pipeline = Pipeline([\n ('imputer', SimpleImputer(strategy=\"median\")),\n ('std_scaler', StandardScaler()),\n])", "_____no_output_____" ], [ "df_numpy = my_pipeline.fit_transform(df)", "_____no_output_____" ], [ "df_numpy \n#Numpy array of df as models will take numpy array as input.", "_____no_output_____" ], [ "df_numpy.shape", "_____no_output_____" ] ], [ [ "## Model Selection", "_____no_output_____" ] ], [ [ "from sklearn.linear_model import LinearRegression\nfrom sklearn.tree import DecisionTreeRegressor\nfrom sklearn.ensemble import RandomForestRegressor\n# model = LinearRegression()\n# model = DecisionTreeRegressor()\nmodel = RandomForestRegressor()\nmodel.fit(df_numpy, df_labels)", "_____no_output_____" ], [ "some_data = df.iloc[:5]", "_____no_output_____" ], [ "some_labels = df_labels.iloc[:5]", "_____no_output_____" ], [ "prepared_data = my_pipeline.transform(some_data)", "_____no_output_____" ], [ "model.predict(prepared_data)", "_____no_output_____" ], [ "list(some_labels)", "_____no_output_____" ] ], [ [ "## Evaluating the model", "_____no_output_____" ] ], [ [ "from sklearn.metrics import mean_squared_error\ndf_predictions = model.predict(df_numpy)\nmse = mean_squared_error(df_labels, df_predictions)\nrmse = np.sqrt(mse)\nrmse", "_____no_output_____" ], [ "# from sklearn.metrics import accuracy_score\n# accuracy_score(some_data, some_labels, normalize=False)", "_____no_output_____" ] ], [ [ "## Cross Validation", "_____no_output_____" ] ], [ [ "from sklearn.model_selection import cross_val_score\nscores = cross_val_score(model, df_numpy, df_labels, scoring=\"neg_mean_squared_error\", cv=10)\nrmse_scores = np.sqrt(-scores)", "_____no_output_____" ], [ "rmse_scores", "_____no_output_____" ], [ "def print_scores(scores):\n print(\"Scores:\", scores)\n print(\"\\nMean:\", scores.mean())\n print(\"\\nStandard deviation:\", scores.std())", "_____no_output_____" ], [ "print_scores(rmse_scores)", "Scores: [2.79289168 2.69441597 4.40018895 2.56972379 3.33073436 2.62687167\n 4.77007351 3.27403209 3.38378214 3.16691711]\n\nMean: 3.3009631251857217\n\nStandard deviation: 0.7076841067486248\n" ] ], [ [ "### Saving Model", "_____no_output_____" ] ], [ [ "from joblib import dump, load\ndump(model, 'final_model.joblib')\ndump(model, 'final_model.sav')", "_____no_output_____" ] ], [ [ "## Testing model on test data", "_____no_output_____" ] ], [ [ "X_test = strat_test_set.drop(\"MEDV\", axis=1)\nY_test = strat_test_set[\"MEDV\"].copy()\nX_test_prepared = my_pipeline.transform(X_test)\nfinal_predictions = model.predict(X_test_prepared)\nfinal_mse = mean_squared_error(Y_test, final_predictions)\nfinal_rmse = np.sqrt(final_mse)", "_____no_output_____" ], [ "final_rmse", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# In-Place Waveform Library Updates\nThis example notebook shows how one can update pulses data in-place without recompiling.\n\n© Raytheon BBN Technologies 2020", "_____no_output_____" ], [ "Set the `SAVE_WF_OFFSETS` flag in order that QGL will output a map of the waveform data within the compiled binary waveform library.", "_____no_output_____" ] ], [ [ "from QGL import *\nimport QGL\nimport os.path\nimport pickle\nQGL.drivers.APS2Pattern.SAVE_WF_OFFSETS = True", "_____no_output_____" ] ], [ [ "Create the usual channel library with a couple of AWGs.", "_____no_output_____" ] ], [ [ "cl = ChannelLibrary(\":memory:\")\nq1 = cl.new_qubit(\"q1\")\naps2_1 = cl.new_APS2(\"BBNAPS1\", address=\"192.168.5.101\") \naps2_2 = cl.new_APS2(\"BBNAPS2\", address=\"192.168.5.102\")\ndig_1 = cl.new_X6(\"X6_1\", address=0)\nh1 = cl.new_source(\"Holz1\", \"HolzworthHS9000\", \"HS9004A-009-1\", power=-30)\nh2 = cl.new_source(\"Holz2\", \"HolzworthHS9000\", \"HS9004A-009-2\", power=-30) \ncl.set_control(q1, aps2_1, generator=h1)\ncl.set_measure(q1, aps2_2, dig_1.ch(1), generator=h2)\ncl.set_master(aps2_1, aps2_1.ch(\"m2\"))\ncl[\"q1\"].measure_chan.frequency = 0e6\ncl.commit()", "Creating engine...\n" ] ], [ [ "Compile a simple sequence.", "_____no_output_____" ] ], [ [ "mf = RabiAmp(cl[\"q1\"], np.linspace(-1, 1, 11))\nplot_pulse_files(mf, time=True)", "Compiled 11 sequences.\n<module 'QGL.drivers.APS2Pattern' from '/Users/growland/workspace/QGL/QGL/drivers/APS2Pattern.py'>\n<module 'QGL.drivers.APS2Pattern' from '/Users/growland/workspace/QGL/QGL/drivers/APS2Pattern.py'>\n" ] ], [ [ "Open the offsets file (in the same directory as the `.aps2` files, one per AWG slice.)", "_____no_output_____" ] ], [ [ "offset_f = os.path.join(os.path.dirname(mf), \"Rabi-BBNAPS1.offsets\")\nwith open(offset_f, \"rb\") as FID:\n offsets = pickle.load(FID)\noffsets", "_____no_output_____" ] ], [ [ "Let's replace every single pulse with a fixed amplitude `Utheta`", "_____no_output_____" ] ], [ [ "pulses = {l: Utheta(q1, amp=0.1, phase=0) for l in offsets}\nwfm_f = os.path.join(os.path.dirname(mf), \"Rabi-BBNAPS1.aps2\")\nQGL.drivers.APS2Pattern.update_wf_library(wfm_f, pulses, offsets)", "_____no_output_____" ] ], [ [ "We see that the data in the file has been updated.", "_____no_output_____" ] ], [ [ "plot_pulse_files(mf, time=True)", "<module 'QGL.drivers.APS2Pattern' from '/Users/growland/workspace/QGL/QGL/drivers/APS2Pattern.py'>\n<module 'QGL.drivers.APS2Pattern' from '/Users/growland/workspace/QGL/QGL/drivers/APS2Pattern.py'>\n" ] ], [ [ "## Profiling \nHow long does this take?", "_____no_output_____" ] ], [ [ "%timeit mf = RabiAmp(cl[\"q1\"], np.linspace(-1, 1, 100))", "Compiled 100 sequences.\nCompiled 100 sequences.\nCompiled 100 sequences.\nCompiled 100 sequences.\nCompiled 100 sequences.\nCompiled 100 sequences.\nCompiled 100 sequences.\nCompiled 100 sequences.\n317 ms ± 6.15 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n" ] ], [ [ "Getting the offsets is fast, and only needs to be done once", "_____no_output_____" ] ], [ [ "def get_offsets():\n offset_f = os.path.join(os.path.dirname(mf), \"Rabi-BBNAPS1.offsets\")\n with open(offset_f, \"rb\") as FID:\n offsets = pickle.load(FID)\n return offsets\n%timeit offsets = get_offsets()", "61.1 µs ± 638 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)\n" ], [ "%timeit pulses = {l: Utheta(q1, amp=0.1, phase=0) for l in offsets}", "39.3 µs ± 1.17 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)\n" ], [ "wfm_f = os.path.join(os.path.dirname(mf), \"Rabi-BBNAPS1.aps2\")\n%timeit QGL.drivers.APS2Pattern.update_wf_library(wfm_f, pulses, offsets)\n# %timeit QGL.drivers.APS2Pattern.update_wf_library(\"/Users/growland/workspace/AWG/Rabi/Rabi-BBNAPS1.aps2\", pulses, offsets)", "1.25 ms ± 19.1 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n" ] ], [ [ "Moral of the story: 300 ms for initial compilation, and roughly 1.3 ms for update_in_place. ", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "empty" ] ] ]

[ "empty" ]

[ [ "empty" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "%load_ext Cython\nimport numpy as np\nnp.set_printoptions(precision=2,suppress=True,linewidth=250,threshold=2000)", "_____no_output_____" ], [ "import numpy as np\nimport pandas as pd\nimport pyBigWig\nimport math\nimport csv\nimport multiprocessing", "_____no_output_____" ], [ "bw = pyBigWig.open(\"/home/musab/bigwig/wgEncodeSydhTfbsHepg2Arid3anb100279IggrabSig.bigWig\")\nchromo = 'chr14'\ntotal_size = bw.chroms()[chromo]\nF=bw.values(chromo,0,total_size,numpy=True)\nF[np.isnan(F)] = 0", "_____no_output_____" ], [ "%%cython -a\n\nimport numpy as np\nimport pandas as pd\nimport math\nimport copy\nimport os\nimport cython\n\n@cython.boundscheck(False)\n@cython.wraparound(False) \ncdef V_BC ( long i, long j,double [::1] CP,double [::1] CiP):\n cdef double P_SUM,iP_SUM\n \n if i == 1:\n P_SUM = CP[j-1]\n iP_SUM = CiP[j-1]\n else:\n P_SUM = CP[j-1]-CP[i-2]\n iP_SUM = CiP[j-1]-CiP[i-2]\n \n try:\n return (P_SUM*(iP_SUM/P_SUM)**2)\n except:\n return 0\n\n@cython.boundscheck(False)\n@cython.wraparound(False) \ncdef mean( long i, long j,double [::1] CP,double [::1] Cip):\n cdef double P_SUM,iP_SUM\n \n if i == 1:\n P_SUM = CP[j-1]\n iP_SUM = Cip[j-1]\n else:\n P_SUM = CP[j-1]-CP[i-2]\n iP_SUM = Cip[j-1]-Cip[i-2]\n \n try:\n return iP_SUM/P_SUM\n except:\n return 0\n \n@cython.boundscheck(False)\n@cython.wraparound(False) \ncdef lookup(long col,long j,long row,double [:,::1] C,double [::1] CP,double [::1] CiP):\n return C[col+j,j] + V_BC(col+(j+1) , row+(j+1),CP,CiP) \n\n@cython.boundscheck(False)\n@cython.wraparound(False) \ncdef SMAWK(long [::1] mri,long [::1] mci, long j,double [:,::1] C,long [:,::1] D,long [::1]mposi,double [::1] CP,double [::1] CiP):\n cdef long [::1] aci\n cdef long [::1] bri\n\n if mci.shape[0] != mri.shape[0]:\n aci=REDUCE(mri,mci,j,C,CP,CiP)\n else:\n aci=mci\n\n if (mri.shape[0]==1 and aci.shape[0]==1):\n C[mri[0]+j+1,j+1]= lookup(aci[0],j,mri[0],C,CP,CiP)\n mposi[mri[0]]=D[mri[0]+j,j]=aci[0]\n\n return \n\n bri = mri[1::2].copy()\n\n SMAWK(bri,aci,j,C,D,mposi,CP,CiP)\n MFILL(mri,aci,j,C,D,mposi,CP,CiP)\n\n@cython.boundscheck(False)\n@cython.wraparound(False) \ncdef REDUCE( long [::1] rows, long [::1] cols, long j,double [:,::1] C,double [::1] CP,double [::1] CiP):\n cdef long p = rows.shape[0]\n cdef long ncols = cols.shape[0]\n cdef long m = cols.shape[0]\n \n predd = np.arange(-1,m+1,dtype=np.int64)\n cdef long [::1] pred = predd\n valuee = np.full((m+1),-1,dtype=np.double)\n cdef double [::1] value = valuee\n\n cdef long a=2 \n cdef long b=1 \n \n rett = np.empty(p,dtype=np.int64)\n cdef long [::1] ret = rett\n cdef long lc = ncols+1\n \n \n value[1]=lookup(cols[0],j,rows[0],C,CP,CiP) \n\n while m > p:\n if value[pred[a]] == -1:\n value[pred[a]] = (lookup(cols[pred[a]-1] ,j, rows[b-1],C,CP,CiP) if cols[pred[a]-1] <= rows[b-1] else 0)\n\n if value[pred[a]] >= (lookup(cols[a-1] ,j, rows[b-1],C,CP,CiP) if cols[a-1] <= rows[b-1] else 0) :\n if b < p :\n b = b+1\n value[a] = (lookup(cols[a-1] ,j, rows[b-1],C,CP,CiP) if cols[a-1] <= rows[b-1] else 0)\n else:\n pred[a+1] = pred[a]\n m = m-1\n\n a = a+1\n\n else:\n pred[a] = pred[pred[a]]\n m = m-1\n if b != 1:\n b = b-1\n else:\n a = a+1\n\n for i in range(p-1,-1,-1):\n ret[i]=cols[pred[lc]-1]\n lc=pred[lc]\n return ret\n\n@cython.boundscheck(False)\n@cython.wraparound(False) \ncdef MFILL( long [::1] ari, long [::1] aci, long j,double [:,::1] C,long [:,::1] D,long [::1]mposi,double [::1] CP,double [::1] CiP):\n cdef long m = ari.shape[0]\n cdef long n = aci.shape[0]\n cdef long ii = n-1\n cdef long start\n\n if (m % 2) == 0:\n start = m-2\n else:\n start = m-1\n\n cdef long s,e,ar,ac,i\n cdef double MAX,cc,vv,CURRENT_MAT\n for i in range(start,-1,-2):\n if (i==0):\n s=int(aci[i])\n e=int(mposi[ari[i+1]])\n\n elif (i==m-1):\n s=int(mposi[ari[i-1]])\n e=int(aci[n-1])\n\n else:\n s=int(mposi[ari[i-1]])\n e=int(mposi[ari[i+1]])\n\n if(e > ari[i]):\n e=int(ari[i]) \n\n MAX = 0\n while True:\n ac = aci[ii]\n ar = ari[i]\n if (ac > ar):\n pass \n else:\n CURRENT_MAT = lookup(ac,j,ar,C,CP,CiP)\n\n if (MAX < CURRENT_MAT):\n MAX = CURRENT_MAT\n mposi[ar]=ac\n C[ar+j+1,j+1]=MAX\n D[ar+j,j]=ac+j\n if(ac<=s or ii==-1):\n break\n ii-=1\n\n@cython.boundscheck(False)\n@cython.wraparound(False) \ncdef findBestK( long n, long k,long [:,::1] D,double [::1] P,double [::1] CP,double [::1] Cip,double [::1] CF):\n cdef double E1 = 0\n cdef double mean1 = mean(1,n,CP,Cip)\n cdef long J,kk,j,a,K,i,t\n for J in range(1,n):\n E1 += (P[J])*(abs(J-mean1))\n\n cdef double bestAlpha = 0\n cdef long bestK = 0\n\n cdef double Dk,Ek,meanK,A,alpha\n cdef long [::1] T\n cdef long [::1] bestT\n \n for kk in range(k,1,-1):\n T = np.zeros(kk+2,dtype=np.int64)\n T[0] = 1\n T[kk+1] = n\n i = n\n for j in range(kk,0,-1):\n T[j] = D[i-1,j]\n i = int(D[i-1,j])\n if i < 1:\n i=1\n\n Dk = mean(T[kk],T[kk+1],CP,Cip)-mean(T[0],T[1],CP,Cip)\n Ek = 1\n\n for K in range(kk+1):\n meanK = mean(T[K],T[K+1],CP,Cip)\n for J in range(T[K],T[K+1]):\n Ek += (P[J])*(abs(J-meanK))\n\n A = 1\n for a in range (kk+1):\n t = T[a]\n A += P[t]\n\n A *= math.sqrt(kk)\n alpha=((((E1/Ek)*Dk)**2)/A) \n\n if alpha > bestAlpha :\n bestK = kk\n bestT = T\n bestAlpha = alpha\n\n cdef double [::1] vol= np.empty(bestK)\n cdef double [::1] leng = np.empty(bestK)\n if bestK != 0:\n for i in range (1,bestK+1):\n vol[i-1]=(CF[bestT[i]]-CF[bestT[i-1]])\n leng[i-1]=(bestT[i]-bestT[i-1])\n else:\n bestT = np.empty(0,dtype=long)\n \n return bestT,bestK,vol,leng\n\n@cython.boundscheck(False)\n@cython.wraparound(False) \ndef L(float [::1] F, long k):\n import matplotlib.pyplot as plt\n import numpy as np\n import pandas as pd\n import math\n import copy\n import time\n import os\n\n \n cdef long n = len(F)\n cdef double F_SUM = sum(F)\n if F_SUM == 0:\n return [],0,[],[]\n\n CC = np.zeros((n+1,k+2),dtype=np.float64)\n DD = np.zeros((n+1,k+2),dtype=np.int64)\n cdef double [:, ::1] C = CC\n cdef long[:, ::1] D = DD \n cdef long sizeMAT \n cdef long[::1] mposi \n\n cdef double [::1] P = np.empty(n)\n cdef double [::1] CP = np.empty(n)\n cdef double [::1] CF = np.empty(n)\n cdef double [::1] CiP = np.empty(n)\n cdef double [::1] Cip = np.empty(n)\n cdef double PP,iP,ip\n cdef double firstValue = F[0]/F_SUM\n CF[0]=(F[0])\n P[0]=(firstValue)\n CP[0]=(firstValue)\n CiP[0]=(firstValue)\n Cip[0]=(firstValue)\n cdef long i,j,dl\n \n for i in range(1,n):\n CF[i]=(CF[i-1]+F[i])\n PP = F[i]/F_SUM\n iP = PP * i\n ip = PP * (i+1)\n P[i]=(PP)\n CP[i]=(CP[i-1]+PP)\n CiP[i]=(CiP[i-1]+iP)\n Cip[i]=(Cip[i-1]+ip)\n \n mminTj = np.zeros(k+2,dtype=np.int64)\n mmaxTj = np.zeros(k+2,dtype=np.int64)\n cdef long[::1] minTj = mminTj\n cdef long[::1] maxTj = mmaxTj\n\n for j in range(0,k+2):\n if j == k+1:\n minTj[j] = n\n else:\n minTj[j] = j\n\n for j in range(0,k+2):\n if j == 0:\n maxTj[j] = 0\n else:\n maxTj[j] = n-k+j-1 \n\n cdef long l,tj,ex=k-2\n cdef double v,f\n cdef long [::1] initial_rc \n for j in range (0,k+1): \n\n if j == 0:\n for tj in range(minTj[j+1],maxTj[j+1]+1+ex):\n C[tj,j+1]=V_BC(1,tj,CP,CiP)\n \n else: \n if (j>=(3)):\n ex-=1\n sizeMAT = n-k+1+ex\n \n if (j != k):\n mposi = np.zeros(sizeMAT-1,dtype=np.int64)\n initial_rc = np.arange(0,sizeMAT-1,dtype=int)\n SMAWK(initial_rc,initial_rc,j,C,D,mposi,CP,CiP)\n else:\n dl = minTj[k]\n for l in range(minTj[k],maxTj[k]+1):\n f = V_BC(l+1,minTj[k+1],CP,CiP)\n v = f + C[l,k]\n if (C[minTj[k+1],k+1] < v ):\n C[minTj[k+1],k+1] = v\n dl = l\n D[maxTj[k],k]=dl\n\n D[n,k+1] = n\n\n cdef long [::1] bestT\n cdef long bestK\n cdef double [::1] vol\n cdef double [::1] leng\n bestT,bestK,vol,leng = findBestK(n,k,D,P,CP,Cip,CF)\n return (bestT,bestK,vol,leng) ", "_____no_output_____" ], [ "def window(tup):\n start = tup[0]\n end = tup[1]\n froM = start\n winSize = 10000\n to = froM+winSize\n '''\n fname = str(str(chromo)+\"-\"+str(start)+\"-\"+str(end)+\".csv\")\n with open(fname, 'w', newline='') as file:\n writer = csv.writer(file)\n writer.writerow([\"START\", \"END\", \"LENGTH\", \"VOLUME\", \"RANGE\"])\n '''\n df = pd.DataFrame(columns=[\"CHR\",\"START\", \"END\", \"LENGTH\", \"VOLUME\", \"RANGE\"])\n while(froM < end):\n try:\n bestT=[]\n bestK=0\n sF=F[froM:to]\n if len(sF) != winSize:\n sF=np.append(sF,np.zeros(winSize - len(sF),dtype=sF.dtype))\n bestT,bestK,vol,leng=(L(sF,6))\n\n '''\n with open(fname, 'a', newline='') as file:\n writer = csv.writer(file)\n for i in range(bestK):\n writer.writerow([froM+bestT[i],froM+bestT[i+1],leng[i],vol[i],str(str(froM)+':'+str(to))])\n '''\n if bestK != 0:\n for i in range(bestK):\n df = df.append({\"CHR\":chromo,\"START\":froM+bestT[i],\"END\":froM+bestT[i+1], \"LENGTH\":leng[i], \"VOLUME\":vol[i], \"RANGE\":str(str(froM)+':'+str(to))},ignore_index=True)\n\n if bestK == 0:\n froM+=winSize\n else:\n froM+=bestT[bestK]\n to = froM+winSize\n\n except Exception as e:\n print(\"From:{}, To:{}, bestK={}\".format(froM,to,bestK))\n print('from Window, ',e)\n froM+=winSize\n to=froM+winSize\n \n return df", "_____no_output_____" ], [ "%%time\n\npath=\"/home/musab/chip-seq/GM12878-H3K27Ac_COPY/\"\nfile = \"wgEncodeBroadHistoneGm12878H3k27acStdSig.bigWig\"\nbw = pyBigWig.open(str(path+file))\nfull_result_list = []\nfor chromo in bw.chroms():\n\n total_size = bw.chroms()[chromo]\n F=bw.values(chromo,0,total_size,numpy=True)\n F[np.isnan(F)] = 0\n\n\n n = multiprocessing.cpu_count()\n frag = math.ceil(total_size/n)\n job = []\n st=0\n while (st<total_size):\n en =st+frag\n job.append((st,en))\n st = en+1\n\n pool = multiprocessing.Pool(processes=n)\n r = pool.map(window,job)\n \n result = pd.concat(r)\n result = result.reset_index(drop=True)\n full_result_list.extend(r)\n\n result.to_csv(str(path+\"results/\"+chromo+'.csv'), index=False)\n pool.close()\n pool.join()\n print(\"Chromo {} Done !\".format(chromo))\n \nfull_result = pd.concat(full_result_list)\nfull_result = full_result.reset_index(drop=True)\nfull_result.to_csv(str(path+\"results/\"+'full.csv'), index=False)", "Chromo chr1 Done !\nChromo chr10 Done !\nChromo chr11 Done !\nChromo chr12 Done !\nChromo chr13 Done !\nChromo chr14 Done !\nChromo chr15 Done !\nChromo chr16 Done !\nChromo chr17 Done !\nChromo chr18 Done !\nChromo chr19 Done !\nChromo chr2 Done !\nChromo chr20 Done !\nChromo chr21 Done !\nChromo chr22 Done !\nChromo chr3 Done !\nChromo chr4 Done !\nChromo chr5 Done !\nChromo chr6 Done !\nChromo chr7 Done !\nChromo chr8 Done !\nChromo chr9 Done !\nChromo chrX Done !\nCPU times: user 46.2 s, sys: 21.6 s, total: 1min 7s\nWall time: 28min 31s\n" ], [ "import os\nimport glob\nimport pandas as pd\nimport math\nos.chdir(\"/home/musab/chip-seq/GM12878-H3K27Ac/results/\")\n\nextension = 'csv'\nall_filenames = [i for i in glob.glob('*.{}'.format(extension))]\n\n#combine all files in the list\ncombined_csv = pd.concat([pd.read_csv(f) for f in all_filenames ])\ncombined_csv = combined_csv.sort_values('VOLUME', ascending=False)\n#export to csv\n# del(combined_csv['RANGE'])\n# del(combined_csv['VOLUME'])\n# del(combined_csv['LENGTH'])\n\n#combined_csv.to_csv( \"annotate.bed\", header=False,index=False, sep='\\t',encoding='utf-8-sig')", "_____no_output_____" ], [ "anno = combined_csv", "_____no_output_____" ], [ "del(anno['RANGE'])\ndel(anno['VOLUME'])\ndel(anno['LENGTH'])", "_____no_output_____" ], [ "length = anno.shape[0]", "_____no_output_____" ], [ "annotate = anno.head(math.ceil(40821*1.5))#math.ceil(length*0.01))", "_____no_output_____" ], [ "annotate.shape", "_____no_output_____" ], [ "annotate.to_csv( \"anno_1.5.bed\", header=False,index=False, sep='\\t',encoding='utf-8-sig')", "_____no_output_____" ], [ "annotate.tail()", "_____no_output_____" ], [ "anno.tail()", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Tutorial 09: Inflows\n\nThis tutorial walks you through the process of introducing inflows of vehicles into a network. Inflows allow us to simulate open networks where vehicles may enter (and potentially exit) the network. This exercise is organized as follows: in section 1 we prepare our inflows variables to support inflows into a merge network supplied by Flow, and in section 2 we simulate the merge network in the presence of inflows.\n\n## 1. Adding Inflows\n\nFor this exercise, we will simulate inflows through a highway network with an on-merge. As we will see, the perturbations caused by vehicles entering through the on-merge leads the formation of congested waves downstream in the main highway.\n\nWe begin by importing the merge scenario class provided by Flow.", "_____no_output_____" ] ], [ [ "from flow.scenarios.merge import MergeScenario", "_____no_output_____" ] ], [ [ "A schematic of the above network is availabe in the figure below. As we can see, the edges at the start of the main highway and the on-merge are named \"inflow_highway\" and \"inflow_merge\" respectively. These names will be important to us when we begin specifying our inflows into the network.\n\n<img src=\"img/merge_scheme.png\" width=\"750\">\n\nWe will also define the types of vehicles that are placed in the network. These types of vehicles will also be of significance to us once the inflows are being defined. For this exercise, we add only one type of vehicle to the network, with the vehicle identifier \"human\":", "_____no_output_____" ] ], [ [ "from flow.core.vehicles import Vehicles\nfrom flow.controllers import IDMController\n\n# create an empty vehicles object\nvehicles = Vehicles()\n\n# add some vehicles to this object of type \"human\"\nvehicles.add(\"human\", \n acceleration_controller=(IDMController, {}),\n speed_mode=\"no_collide\", # we use the speed mode \"no_collide\" for better dynamics at the merge\n num_vehicles=20)", "_____no_output_____" ] ], [ [ "Next, we are ready to import and create an empty inflows object.", "_____no_output_____" ] ], [ [ "from flow.core.params import InFlows\n\ninflow = InFlows()", "_____no_output_____" ] ], [ [ "The `InFlows` object is provided as an input during the scenario creation process via the `NetParams` parameter. Introducing these inflows into the network is handled by the backend scenario generation processes during instantiation of the scenario object.\n\nIn order to add new inflows of vehicles of pre-defined types onto specific edges and lanes in the network, we use the `InFlows` object's `add` method. This function accepts the following parameters:\n\n* **veh_type**: type of vehicles entering the edge, must match one of the types set in the Vehicles class\n* **edge**: starting edge for vehicles in this inflow, must match an edge name in the network\n* **veh_per_hour**: number of vehicles entering from the edge per hour, may not be achievable due to congestion and safe driving behavior\n* other parameters, including: **start**, **end**, and **probability**. See documentation for more information.\n\nIn addition to the above parameters, several optional inputs to the `add` method may be found within sumo's documentation at: http://sumo.dlr.de/wiki/Definition_of_Vehicles,_Vehicle_Types,_and_Routes. Some important features include:\n\n* **departLane**: specifies which lane vehicles will enter from on the edge, may be specified as \"all\" or \"random\"\n* **departSpeed**: speed of the vehicles once they enter the network\n\nWe begin by adding inflows of vehicles at a rate of 2000 veh/hr through *all* lanes on the main highways as follows:", "_____no_output_____" ] ], [ [ "inflow.add(veh_type=\"human\",\n edge=\"inflow_highway\",\n vehs_per_hour=2000,\n departSpeed=10,\n departLane=\"random\")", "_____no_output_____" ] ], [ [ "Next, we specify a second inflow of vehicles through the on-merge lane at a rate of only 100 veh/hr.", "_____no_output_____" ] ], [ [ "inflow.add(veh_type=\"human\",\n edge=\"inflow_merge\",\n vehs_per_hour=100,\n departSpeed=10,\n departLane=\"random\")", "_____no_output_____" ] ], [ [ "## 2. Running Simulations with Inflows\n\nWe are now ready to test our inflows in simulation. As mentioned in section 1, the inflows are specified in the `NetParams` object, in addition to all other network-specific parameters. For the merge network, this is done as follows: ", "_____no_output_____" ] ], [ [ "from flow.scenarios.merge import ADDITIONAL_NET_PARAMS\nfrom flow.core.params import NetParams\n\nadditional_net_params = ADDITIONAL_NET_PARAMS.copy()\n\n# we choose to make the main highway slightly longer\nadditional_net_params[\"pre_merge_length\"] = 500\n\nnet_params = NetParams(inflows=inflow, # our inflows\n no_internal_links=False,\n additional_params=additional_net_params)", "_____no_output_____" ] ], [ [ "Finally, we execute the simulation following simulation creation techniques we learned from exercise 1 using the below code block. Running this simulation, we see an excessive number of vehicles entering from the main highway, but only a sparse number of vehicles entering from the on-merge. Nevertheless, this volume of merging vehicles is sufficient to form congestive patterns within the main highway.", "_____no_output_____" ] ], [ [ "from flow.core.params import SumoParams, EnvParams, InitialConfig\nfrom flow.envs.loop.loop_accel import AccelEnv, ADDITIONAL_ENV_PARAMS\nfrom flow.core.experiment import SumoExperiment\n\nsumo_params = SumoParams(render=True,\n sim_step=0.2)\n\nenv_params = EnvParams(additional_params=ADDITIONAL_ENV_PARAMS)\n\ninitial_config = InitialConfig()\n\nscenario = MergeScenario(name=\"merge-example\",\n vehicles=vehicles,\n net_params=net_params,\n initial_config=initial_config)\n\nenv = AccelEnv(env_params, sumo_params, scenario)\n\nexp = SumoExperiment(env, scenario)\n\n_ = exp.run(1, 1500)", "**********************************************************\n**********************************************************\n**********************************************************\nWARNING: Inflows will cause computational performance to\nsignificantly decrease after large number of rollouts. In \norder to avoid this, set SumoParams(restart_instance=True).\n**********************************************************\n**********************************************************\n**********************************************************\nRound 0, return: 461.69990202420087\nAverage, std return: 461.69990202420087, 0.0\nAverage, std speed: 4.068107238565223, 0.0\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "%matplotlib inline\nfrom IPython.core.display import display, HTML\ndisplay(HTML(\"<style>.container { width:100% !important; }</style>\"))\n", "_____no_output_____" ], [ "import numpy as np\nnp.set_printoptions(precision=3, suppress=True)\nimport library.helpers as h\nimport matplotlib.pyplot as plt\nimport numpy as np\nfrom sklearn.metrics import auc\nfrom scipy.integrate import simps\nfrom scipy.interpolate import interp1d\n\n\nLOG_NAME = \"unix_forensic\"\nVIZUALIZATIONS_DIR = \"visualizations\"\n\nfig = plt.figure(figsize=(10,10))\n\n# TODO redo this one [ \nMODEL_NAMES = [ \"lstm-ae\", \"triplet-la\", \"triplet-jd-la-all-40-70-all-all\", \"triplet-jd-la-non-60-65-non-non\", \"triplet-jd-la-00500-60-65-all-all-00001\", \n \"triplet-jd-la-00500-60-65-all-00001-non\", \"triplet-jd-la-01000-60-65-all-all-00001\", \"triplet-jd-la-05000-60-65-all-all-00001\", \"jd-la-x05000-pt60-nt65-llall-lcall-ee00001-ep20\", \"jd-la-xnon-pt40-nt70-llall-lcall-ee00001-ep50\"] \n\n\nMODEL_NAMES_JD = [\n \"triplet-jd-la-all-40-70-all-all\", # baseline, all labels\n \"triplet-jd-la-non-60-65-non-non\",\"triplet-jd-la-non-40-70-non-non\", # jd no labels, 10 ep\n \"jd-la-xnon-pt60-nt65-llnon-lcnon-eenon-ep50\", \n \"jd-la-xnon-pt30-nt75-llnon-lcnon-eenon-ep50\"\n #\"jd-la-x02000-pt55-nt65-llnon-lc00003-ee00002-ep10\",\"jd-la-x02000-pt55-nt65-llnon-lc00003-ee00002-ep20\",\n #\"triplet-jd-la-05000-60-65-all-all-00001\", \"jd-la-x05000-pt60-nt65-llall-lcall-ee00001-ep20\",\n] \n\nMODEL_NAMES_LA = [\n \"triplet-jd-la-all-40-70-all-all\", # baseline, all labels\n \"jd-la-x00500-pt30-nt70-llall-lcall-ee00002-ep30\", # 500 labels \n \"jd-la-x01000-pt30-nt70-llall-lcall-ee00002-ep30\", # 1000 labels\n \"jd-la-x02000-pt30-nt70-llall-lcall-ee00002-ep30\", # 2000 labels\n \"jd-la-x02500-pt30-nt70-llall-lcall-ee00002-ep30\", # 2500 labels \n \"jd-la-x05000-pt30-nt70-llall-lcall-ee00002-ep30\", # 5000 labels \n] \n\nMODEL_NAMES_NT60 = [\n \"jd-la-xnon-pt10-nt60-llnon-lcnon-ee00002-ep30\", # \n \"jd-la-xnon-pt20-nt60-llnon-lcnon-ee00002-ep30\", # \n \"jd-la-xnon-pt30-nt60-llnon-lcnon-ee00002-ep30\", # \n \"jd-la-xnon-pt40-nt60-llnon-lcnon-ee00002-ep30\", # \n \"jd-la-xnon-pt50-nt60-llnon-lcnon-ee00002-ep30\",\n]\n\nMODEL_NAMES_NT70 = [\n \"jd-la-xnon-pt10-nt70-llnon-lcnon-ee00002-ep30\", # \n \"jd-la-xnon-pt20-nt70-llnon-lcnon-ee00002-ep30\", # \n \"jd-la-xnon-pt30-nt70-llnon-lcnon-ee00002-ep30\", # \n \"jd-la-xnon-pt40-nt70-llnon-lcnon-ee00002-ep30\", # \n \"jd-la-xnon-pt50-nt70-llnon-lcnon-ee00002-ep30\",\n \"jd-la-xnon-pt60-nt70-llnon-lcnon-ee00002-ep30\"\n]\n\nMODEL_NAMES_NT80 = [\n \"jd-la-xnon-pt10-nt80-llnon-lcnon-ee00002-ep30\", # \n \"jd-la-xnon-pt20-nt80-llnon-lcnon-ee00002-ep30\", # \n \"jd-la-xnon-pt30-nt80-llnon-lcnon-ee00002-ep30\", # \n \"jd-la-xnon-pt40-nt80-llnon-lcnon-ee00002-ep30\", # \n \"jd-la-xnon-pt50-nt80-llnon-lcnon-ee00002-ep30\",\n \"jd-la-xnon-pt60-nt80-llnon-lcnon-ee00002-ep30\",\n \"jd-la-xnon-pt70-nt80-llnon-lcnon-ee00002-ep30\"\n]\n\nMODEL_NAMES_10 = [\n #\"jd-la-xnon-pt10-nt90-llnon-lcnon-ee00002-ep20\",\n #\"jd-la-xnon-pt20-nt90-llnon-lcnon-ee00002-ep20\",\n #\"jd-la-xnon-pt30-nt90-llnon-lcnon-ee00002-ep20\",\n #\"jd-la-xnon-pt40-nt90-llnon-lcnon-ee00002-ep20\",\n \"jd-la-xnon-pt50-nt90-llnon-lcnon-ee00002-ep20\",\n \"jd-la-xnon-pt52-nt70-llnon-lcnon-ee00002-ep30\",\n #\"jd-la-xnon-pt60-nt90-llnon-lcnon-ee00002-ep20\",\n #\"jd-la-xnon-pt70-nt90-llnon-lcnon-ee00002-ep20\",\n #\"jd-la-xnon-pt80-nt90-llnon-lcnon-ee00002-ep20\",\n \"jd-la-xnon-pt52-nt90-llnon-lcnon-ee00002-ep200\",\n \n]\n\n\n\nMODEL_NAMES = MODEL_NAMES_LA\n\n# \"triplet-jd-la-2000-55-70-40\", \"lstm-ae\", \"triplet-la\", \"triplet-jd-la-3000-55-70-55-30\", \"triplet-jd-la-ma-1500\"\n# [\"triplet-jd-la-2000-55-70-40\",\"triplet-jd-la-3000-55-70-55-30\",\"triplet-jd-la-3000-50-70-55-30\"]\n# [ \"triplet-jd-la-1500-055-70-40\", \"triplet-jd-la-1500-55-75-40\"]\n# \"triplet-jd-la-1500-060-70-40\", \"triplet-jd-la-1500-055-70-40\", \"triplet-jd-la-1500-050-70-40\",\n# [\"triplet-jd-la-1500-050-065-25\", \"triplet-jd-la-1500-065-070-25\", \"triplet-jd-la-1500-060-060-25\"]\n\n#[\"triplet-jd-la-ma-500-02-03\",\"triplet-jd-la-ma-750\",\"triplet-jd-la-ma-1000\", # \"#4169e1\",\n# \"triplet-jd-la-ma-1500\",\n# \"triplet-jaccard\",\"triplet-jaccard-margin\",\n# \"triplet-label\", # all labels\n# \"lstm-ae\"]\nCOLORS = h.get_N_HexCol(len(MODEL_NAMES)+1)\n\nfac_n = np.arange(0, 1.0, 0.0005)\n\nbaseline_valid_accepts = h.load_from_json(\"data/ji_%s_basline-jaccard_valid.json\"%LOG_NAME)\nbaseline_false_accepts = [np.round(f,5) for f in h.load_from_json(\"data/ji_%s_basline-jaccard_false.json\"%LOG_NAME)]\n\ninterpolated_vac = interp1d(baseline_false_accepts, baseline_valid_accepts)\nvac_n = interpolated_vac(fac_n)\nauc_score = auc(fac_n, vac_n)\n\n\nplt.plot( fac_n, interpolated_vac(fac_n), color='r', label=\"%0.3f,baseline\"%auc_score)\nplt.xscale(\"log\")\n\n\nfor i, model_name in enumerate(MODEL_NAMES):\n #print(model_name)\n valid_accepts = h.load_from_json(\"data/%s_%s_valid.json\"%(model_name, LOG_NAME))\n false_accepts = h.load_from_json(\"data/%s_%s_false.json\"%(model_name, LOG_NAME))\n interpolated_vac = interp1d(false_accepts, valid_accepts)\n auc_score = auc(fac_n, interpolated_vac(fac_n))\n plt.plot( fac_n, interpolated_vac(fac_n) , color=COLORS[i+1], label=\"%0.3f, %s\"%(auc_score, model_name))\n\nplt.title(\"VAR/FAR (%s) LA\"%LOG_NAME)\nplt.xlabel('FAR')\nplt.ylabel('VAL')\nplt.legend(loc='lower right')\nplt.show()\n#plt.savefig(\"%s/roc_%s.png\"%(VIZUALIZATIONS_DIR, LOG_NAME))\n", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# IPython magics\n\nThis notebook is used for testing nbqa with ipython magics.", "_____no_output_____" ] ], [ [ "from random import randint\nfrom IPython import get_ipython", "_____no_output_____" ] ], [ [ "## Cell magics", "_____no_output_____" ] ], [ [ "%%bash\n\nfor n in {1..10}\ndo\n echo -n \"$n \"\ndone", "1 2 3 4 5 6 7 8 9 10 " ], [ "%%time\n\nimport operator\n\n\ndef compute(operand1,operand2, bin_op):\n \"\"\"Perform input binary operation over the given operands.\"\"\"\n return bin_op(operand1, operand2)\n\n\ncompute(5,1, operator.add)", "CPU times: user 31 µs, sys: 4 µs, total: 35 µs\nWall time: 37.9 µs\n" ] ], [ [ "## Help Magics", "_____no_output_____" ] ], [ [ "str.split??", "\u001b[0;31mSignature:\u001b[0m \u001b[0mstr\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msplit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m/\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msep\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmaxsplit\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;31mDocstring:\u001b[0m\nReturn a list of the words in the string, using sep as the delimiter string.\n\nsep\n The delimiter according which to split the string.\n None (the default value) means split according to any whitespace,\n and discard empty strings from the result.\nmaxsplit\n Maximum number of splits to do.\n -1 (the default value) means no limit.\n\u001b[0;31mType:\u001b[0m method_descriptor\n" ], [ "# would this comment also be considered as magic?\nstr.split?", "\u001b[0;31mSignature:\u001b[0m \u001b[0mstr\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msplit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m/\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msep\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmaxsplit\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;31mDocstring:\u001b[0m\nReturn a list of the words in the string, using sep as the delimiter string.\n\nsep\n The delimiter according which to split the string.\n None (the default value) means split according to any whitespace,\n and discard empty strings from the result.\nmaxsplit\n Maximum number of splits to do.\n -1 (the default value) means no limit.\n\u001b[0;31mType:\u001b[0m method_descriptor\n" ], [ " ?str.splitlines", "\u001b[0;31mSignature:\u001b[0m \u001b[0mstr\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msplitlines\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m/\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mkeepends\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;31mDocstring:\u001b[0m\nReturn a list of the lines in the string, breaking at line boundaries.\n\nLine breaks are not included in the resulting list unless keepends is given and\ntrue.\n\u001b[0;31mType:\u001b[0m method_descriptor\n" ] ], [ [ "## Shell magics", "_____no_output_____" ] ], [ [ " !grep -r '%%HTML' . | wc -l", "2\n" ], [ "flake8_version = !pip list 2>&1 | grep flake8\n\nif flake8_version:\n print(flake8_version)", "['flake8 3.8.4']\n" ] ], [ [ "## Line magics", "_____no_output_____" ] ], [ [ " %time randint(5,10)", "CPU times: user 77 µs, sys: 9 µs, total: 86 µs\nWall time: 201 µs\n" ], [ "if __debug__:\n %time compute(5,1, operator.mul)", "CPU times: user 6 µs, sys: 0 ns, total: 6 µs\nWall time: 9.54 µs\n" ], [ "%time get_ipython().run_line_magic(\"lsmagic\", \"\")", "CPU times: user 102 µs, sys: 13 µs, total: 115 µs\nWall time: 124 µs\n" ], [ "import pprint\nimport sys\n\n%time pretty_print_object = pprint.PrettyPrinter(\\\n indent=4, width=80, stream=sys.stdout, compact=True, depth=5\\\n )", "CPU times: user 29 µs, sys: 0 ns, total: 29 µs\nWall time: 33.4 µs\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import numpy as np\nimport pandas as pd\nimport matplotlib.pyplot as plt\nimport pystan\nimport pybaseball\nimport arviz as az", "_____no_output_____" ], [ "player_names = [\"Peter Alonso\",\"Keston Hiura\",\"Fernando Tatis Jr.\",\"Harold Ramirez\",\"Jose Trevino\",\"Yordan Alvarez\",\"Vladimir Guerrero Jr.\",\"Steve Wilkerson\"]", "_____no_output_____" ], [ "batter_data = pybaseball.batting_stats(2019)", "_____no_output_____" ], [ "batter_data = batter_data.loc[batter_data['Name'].isin(player_names)].set_index(\"Name\")\nbatter_data", "_____no_output_____" ], [ "binomial_data = {}\nfor player in batter_data.index:\n hits = int(batter_data.loc[player][\"H\"])\n ab_minus_hits = int(batter_data.loc[player][\"AB\"]) - hits\n ab_outcomes = [0]*ab_minus_hits + [1]*hits\n binomial_data.update({player:ab_outcomes})\nat_bat_totals = {}\nfor player in batter_data.index:\n at_bat_count = int(batter_data.loc[player][\"AB\"])\n at_bat_totals.update({player:at_bat_count})\nhit_totals = {}\nfor player in batter_data.index:\n hit_count = int(batter_data.loc[player][\"H\"])\n hit_totals.update({player:hit_count})", "_____no_output_____" ] ], [ [ "$BA_i \\sim Beta(81,219)$\n\n$y_i \\sim Bin(AB_i,BA_i)$\n\n$i=1,2,...,8$", "_____no_output_____" ] ], [ [ "#https://mc-stan.org/users/documentation/case-studies/rstan_workflow.html\n#https://people.duke.edu/~ccc14/sta-663/PyStan.html\n#http://varianceexplained.org/statistics/beta_distribution_and_baseball/\n\nmodel_code = '''\ndata {\n int<lower=0> N;\n int<lower=0> at_bats[N];\n int<lower=0> hits[N];\n real<lower=0> A;\n real<lower=0> B;\n}\nparameters {\n real<lower=0,upper=1> AVG[N];\n}\nmodel {\n AVG ~ beta(A, B);\n hits ~ binomial(at_bats, AVG);\n}\ngenerated quantities {\n vector[N] log_lik;\n vector[N] predicted_hits;\n for (i in 1:N) {\n log_lik[i] = binomial_lpmf(hits[i] | at_bats[i], AVG[i]);\n predicted_hits[i] = binomial_rng(at_bats[i], AVG[i]);\n }\n}\n'''\n\nmodel_data = dict(N=8, hits=list(hit_totals.values()),at_bats=list(at_bat_totals.values()),A=81,B=219)\nstan_model = pystan.StanModel(model_code=model_code)\nfit = stan_model.sampling(data=model_data)\n", "INFO:pystan:COMPILING THE C++ CODE FOR MODEL anon_model_4d40fdfa42b9a644c6433b2a714e9368 NOW.\n" ], [ "print(fit)", "Inference for Stan model: anon_model_4d40fdfa42b9a644c6433b2a714e9368.\n4 chains, each with iter=2000; warmup=1000; thin=1; \npost-warmup draws per chain=1000, total post-warmup draws=4000.\n\n mean se_mean sd 2.5% 25% 50% 75% 97.5% n_eff Rhat\nAVG[1] 0.26 1.7e-4 0.01 0.24 0.25 0.26 0.27 0.29 7260 1.0\nAVG[2] 0.29 2.2e-4 0.02 0.26 0.28 0.29 0.3 0.33 6842 1.0\nAVG[3] 0.3 2.2e-4 0.02 0.26 0.28 0.29 0.31 0.33 6528 1.0\nAVG[4] 0.29 2.1e-4 0.02 0.25 0.27 0.29 0.3 0.32 7373 1.0\nAVG[5] 0.27 2.1e-4 0.02 0.24 0.26 0.27 0.28 0.3 5894 1.0\nAVG[6] 0.27 2.0e-4 0.02 0.24 0.26 0.27 0.28 0.31 6362 1.0\nAVG[7] 0.27 2.7e-4 0.02 0.23 0.25 0.27 0.28 0.31 5854 1.0\nAVG[8] 0.25 2.1e-4 0.02 0.22 0.24 0.25 0.26 0.28 6435 1.0\nlog_lik[1] -3.64 0.01 0.49 -5.01 -3.74 -3.45 -3.33 -3.29 1873 1.0\nlog_lik[2] -3.62 0.01 0.7 -5.54 -3.9 -3.38 -3.12 -3.03 3387 1.0\nlog_lik[3] -3.71 0.01 0.73 -5.63 -4.01 -3.47 -3.18 -3.06 4215 1.0\nlog_lik[4] -3.48 0.01 0.58 -5.12 -3.67 -3.26 -3.07 -3.02 2921 1.0\nlog_lik[5] -3.49 0.01 0.44 -4.69 -3.58 -3.33 -3.21 -3.18 1371 1.0\nlog_lik[6] -3.42 9.4e-3 0.39 -4.54 -3.52 -3.27 -3.17 -3.14 1673 1.0\nlog_lik[7] -2.64 5.1e-3 0.22 -3.24 -2.7 -2.56 -2.5 -2.49 1877 1.0\nlog_lik[8] -3.6 0.01 0.73 -5.56 -3.88 -3.36 -3.06 -2.95 3949 1.0\npredicted_hits[1] 157.05 0.2 13.74 131.0 148.0 157.0 166.0 184.0 4592 1.0\npredicted_hits[2] 91.42 0.14 10.18 72.0 84.0 91.0 98.0 111.0 4960 1.0\npredicted_hits[3] 98.44 0.15 10.08 79.0 92.0 98.0 105.0 119.0 4683 1.0\npredicted_hits[4] 90.08 0.15 9.96 71.0 83.0 90.0 97.0 110.0 4394 1.0\npredicted_hits[5] 125.81 0.18 12.17 103.0 118.0 125.0 134.0 151.0 4739 1.0\npredicted_hits[6] 115.5 0.18 11.65 93.0 108.0 115.0 123.0 139.0 4335 1.0\npredicted_hits[7] 32.09 0.08 5.42 22.0 28.0 32.0 36.0 43.0 4109 1.0\npredicted_hits[8] 81.14 0.14 9.27 64.0 75.0 81.0 87.0 100.0 4561 1.0\nlp__ -3107 0.05 1.95 -3111 -3108 -3107 -3105 -3104 1748 1.0\n\nSamples were drawn using NUTS at Thu Nov 21 20:41:52 2019.\nFor each parameter, n_eff is a crude measure of effective sample size,\nand Rhat is the potential scale reduction factor on split chains (at \nconvergence, Rhat=1).\n" ], [ "prior_model_code = '''\ndata {\n int<lower=0> N;\n real<lower=0> A;\n real<lower=0> B;\n}\n\nparameters {\n real<lower=0,upper=1> AVG[N];\n}\n\nmodel {\n AVG ~ beta(A, B);\n }\n'''\n\nprior_model_data = dict(N=8,A=81,B=219)\nstan_model_prior = pystan.StanModel(model_code=prior_model_code)\nprior_fit = stan_model_prior.sampling(data=prior_model_data)", "INFO:pystan:COMPILING THE C++ CODE FOR MODEL anon_model_390c11193c42041ad1d6bfb9bf37ebfd NOW.\n" ], [ "stan_data = az.from_pystan(posterior=fit,\n prior=prior_fit,\n observed_data=\"hits\",\n posterior_predictive=\"predicted_hits\",\n log_likelihood=\"log_lik\",\n posterior_model=stan_model,\n coords={\"player\":list(hit_totals.keys())},\n dims={'AVG': ['player'], 'hits': ['player'], 'log_lik': ['player'], 'predicted_hits': ['player']})", "_____no_output_____" ], [ "density_plots = az.plot_density([stan_data.posterior,stan_data.prior],data_labels=[\"Posterior\",\"Prior\"])", "_____no_output_____" ], [ "az.plot_ppc(stan_data, data_pairs = {\"hits\" : \"predicted_hits\"},flatten=[\"player\"])", "_____no_output_____" ] ] ]

[ "code", "markdown", "code" ]

[ [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Introduction to Pandas", "_____no_output_____" ] ], [ [ "import pandas\npandas.__version__", "_____no_output_____" ], [ "import pandas as pd", "_____no_output_____" ] ] ]

[ "markdown", "code" ]

[ [ "markdown" ], [ "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "[](https://colab.research.google.com/github/ourownstory/neural_prophet/blob/master/example_notebooks/sub_daily_data_yosemite_temps.ipynb)", "_____no_output_____" ], [ "# Sub-daily data\nNeuralProphet can make forecasts for time series with sub-daily observations by passing in a dataframe with timestamps in the ds column. The format of the timestamps should be `YYYY-MM-DD HH:MM:SS` - see the example csv [here](https://github.com/ourownstory/neural_prophet/blob/master/example_data/yosemite_temps.csv). When sub-daily data are used, daily seasonality will automatically be fit. \n\nHere we fit NeuralProphet to data with 5-minute resolution (daily temperatures at Yosemite). ", "_____no_output_____" ] ], [ [ "if 'google.colab' in str(get_ipython()):\n !pip install git+https://github.com/ourownstory/neural_prophet.git # may take a while\n #!pip install neuralprophet # much faster, but may not have the latest upgrades/bugfixes\n data_location = \"https://raw.githubusercontent.com/ourownstory/neural_prophet/master/\"\nelse:\n data_location = \"../\"", "_____no_output_____" ], [ "import pandas as pd\nfrom neuralprophet import NeuralProphet, set_log_level\n# set_log_level(\"ERROR\")\ndf = pd.read_csv(data_location + \"example_data/yosemite_temps.csv\")", "_____no_output_____" ] ], [ [ "Now we will attempt to forecast the next 7 days. The `5min` data resulution means that we have `60/5*24=288` daily values. Thus, we want to forecast `7*288` periods ahead.\n\nUsing some common sense, we set:\n* First, we disable weekly seasonality, as nature does not follow the human week's calendar.\n* Second, we disable changepoints, as the dataset only contains two months of data", "_____no_output_____" ] ], [ [ "m = NeuralProphet(\n n_changepoints=0,\n weekly_seasonality=False,\n)\nmetrics = m.fit(df, freq='5min')\nfuture = m.make_future_dataframe(df, periods=7*288, n_historic_predictions=len(df))\nforecast = m.predict(future)\nfig = m.plot(forecast)\n# fig_comp = m.plot_components(forecast)\nfig_param = m.plot_parameters()", "INFO - (NP.forecaster._handle_missing_data) - 12 NaN values in column y were auto-imputed.\nINFO - (NP.utils.set_auto_seasonalities) - Disabling yearly seasonality. Run NeuralProphet with yearly_seasonality=True to override this.\nINFO - (NP.config.set_auto_batch_epoch) - Auto-set batch_size to 128\nINFO - (NP.config.set_auto_batch_epoch) - Auto-set epochs to 6\n" ] ], [ [ "The daily seasonality seems to make sense, when we account for the time being recorded in GMT, while Yosemite local time is GMT-8.\n\n## Improving trend and seasonality \nAs we have `288` daily values recorded, we can increase the flexibility of `daily_seasonality`, without danger of overfitting. \n\nFurther, we may want to re-visit our decision to disable changepoints, as the data clearly shows changes in trend, as is typical with the weather. We make the following changes:\n* increase the `changepoints_range`, as the we are doing a short-term prediction\n* inrease the `n_changepoints` to allow to fit to the sudden changes in trend\n* carefully regularize the trend changepoints by setting `trend_reg` in order to avoid overfitting", "_____no_output_____" ] ], [ [ "m = NeuralProphet(\n changepoints_range=0.95,\n n_changepoints=50,\n trend_reg=1.5,\n weekly_seasonality=False,\n daily_seasonality=10,\n)\nmetrics = m.fit(df, freq='5min')\nfuture = m.make_future_dataframe(df, periods=60//5*24*7, n_historic_predictions=len(df))\nforecast = m.predict(future)\nfig = m.plot(forecast)\n# fig_comp = m.plot_components(forecast)\nfig_param = m.plot_parameters()", "INFO - (NP.config.__post_init__) - Note: Trend changepoint regularization is experimental.\nINFO - (NP.forecaster._handle_missing_data) - 12 NaN values in column y were auto-imputed.\nINFO - (NP.utils.set_auto_seasonalities) - Disabling yearly seasonality. Run NeuralProphet with yearly_seasonality=True to override this.\nINFO - (NP.config.set_auto_batch_epoch) - Auto-set batch_size to 128\nINFO - (NP.config.set_auto_batch_epoch) - Auto-set epochs to 6\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Deep Deterministic Policy Gradients (DDPG)\n---\nIn this notebook, we train DDPG with OpenAI Gym's Pendulum-v0 environment.\n\n### 1. Import the Necessary Packages", "_____no_output_____" ] ], [ [ "import gym\nimport random\nimport torch\nimport numpy as np\nfrom collections import deque\nimport matplotlib.pyplot as plt\n%matplotlib inline\n\nfrom ddpg_agent import Agent", "_____no_output_____" ] ], [ [ "### 2. Instantiate the Environment and Agent", "_____no_output_____" ] ], [ [ "env = gym.make('Pendulum-v0')\nenv.seed(2)\nagent = Agent(state_size=3, action_size=1, random_seed=2)", "_____no_output_____" ] ], [ [ "\nObservation\nType: Box(3)\n\n|Num \t|Observation \t|Min| \tMax|\n| --- | --- | --- |--- |\n0 |\tcos(theta) \t| -1.0 \t|1.0|\n1 |\tsin(theta) \t|-1.0 \t|1.0|\n2 |\ttheta dot \t|-8.0 \t|8.0|\n\nActions\nType: Box(1)\n\n| Num \t| Action | \tMin \t| Max| \n| --- | --- | --- |--- |\n| 0 \t| Joint effort \t| -2.0 \t| 2.0| ", "_____no_output_____" ], [ "### 3. Train the Agent with DDPG", "_____no_output_____" ] ], [ [ "def ddpg(n_episodes=100, max_t=300, print_every=100):\n scores_deque = deque(maxlen=print_every)\n scores = []\n for i_episode in range(1, n_episodes+1):\n state = env.reset()\n agent.reset()\n score = 0\n for t in range(max_t):\n action = agent.act(state)\n next_state, reward, done, _ = env.step(action)\n agent.step(state, action, reward, next_state, done)\n state = next_state\n score += reward\n if done:\n break \n scores_deque.append(score)\n scores.append(score)\n print('\\rEpisode {}\\tAverage Score: {:.2f}'.format(i_episode, np.mean(scores_deque)), end=\"\")\n torch.save(agent.actor_local.state_dict(), 'checkpoint_actor.pth')\n torch.save(agent.critic_local.state_dict(), 'checkpoint_critic.pth')\n if i_episode % print_every == 0:\n print('\\rEpisode {}\\tAverage Score: {:.2f}'.format(i_episode, np.mean(scores_deque)))\n \n return scores\n\nscores = ddpg()\n\nfig = plt.figure()\nax = fig.add_subplot(111)\nplt.plot(np.arange(1, len(scores)+1), scores)\nplt.ylabel('Score')\nplt.xlabel('Episode #')\nplt.show()", "Episode 100\tAverage Score: -595.74\n" ] ], [ [ "### 4. Watch a Smart Agent!", "_____no_output_____" ] ], [ [ "agent.actor_local.load_state_dict(torch.load('checkpoint_actor.pth'))\nagent.critic_local.load_state_dict(torch.load('checkpoint_critic.pth'))\n\nstate = env.reset()\nfor t in range(500):\n action = agent.act(state, add_noise=False)\n env.render()\n state, reward, done, _ = env.step(action)\n if done:\n break \n\nenv.close()", "_____no_output_____" ] ], [ [ "### 6. Explore\n\nIn this exercise, we have provided a sample DDPG agent and demonstrated how to use it to solve an OpenAI Gym environment. To continue your learning, you are encouraged to complete any (or all!) of the following tasks:\n- Amend the various hyperparameters and network architecture to see if you can get your agent to solve the environment faster than this benchmark implementation. Once you build intuition for the hyperparameters that work well with this environment, try solving a different OpenAI Gym task!\n- Write your own DDPG implementation. Use this code as reference only when needed -- try as much as you can to write your own algorithm from scratch.\n- You may also like to implement prioritized experience replay, to see if it speeds learning. \n- The current implementation adds Ornsetein-Uhlenbeck noise to the action space. However, it has [been shown](https://blog.openai.com/better-exploration-with-parameter-noise/) that adding noise to the parameters of the neural network policy can improve performance. Make this change to the code, to verify it for yourself!\n- Write a blog post explaining the intuition behind the DDPG algorithm and demonstrating how to use it to solve an RL environment of your choosing. ", "_____no_output_____" ] ], [ [ "int(1e5)", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import nglview as nv\nfrom nglview.utils import get_colors_from_b64\n\nview = nv.demo()\nview", "_____no_output_____" ], [ "view.add_surface(opacity=0.4)", "_____no_output_____" ], [ "view.render_image()", "_____no_output_____" ], [ "c0 = get_colors_from_b64(view._image_data)", "_____no_output_____" ], [ "view.color_by('residueindex')", "_____no_output_____" ], [ "view.render_image()", "_____no_output_____" ], [ "c1 = get_colors_from_b64(view._image_data)", "_____no_output_____" ], [ "assert c0 != c1", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "DOC" ]

[ "DOC" ]

[ "DOC" ]

[ [ [ "\nprices_lookup = {'apple':2.88, 'oranges':3.56, 'milk':6.12}\n\nprint('precio de las manzanas {:<10.5f} €'.format(prices_lookup['apple']))\n # precio de las manzanas 2.88000 €\n\nprices_lookup # {'apple': 2.88, 'oranges': 3.56, 'milk': 6.12}\n", "precio de las manzanas 2.88000 €\n" ], [ "d = {'numbers':123,'list':[1,2,3],'dict':{'nombre':'david','apellido':'martin'}}\n\nd['numbers'] # 123\nd['list'][0] #1\nprint('me llamo {} {} '.format(d['dict']['nombre'], d['dict']['apellido'])) # me llamo david martin \n", "123\n1\nme llamo david martin \n" ], [ "prices_lookup = {'apple':2.88, 'oranges':3.56, 'milk':6.12}\n\nprices_lookup['melon'] = 5.86\n\nprices_lookup['apple']= 3.30\n\nprices_lookup # {'apple': 3.3, 'oranges': 3.56, 'milk': 6.12, 'melon': 5.86}\n\ndel prices_lookup['melon']\n\nprices_lookup # {'apple': 3.3, 'oranges': 3.56, 'milk': 6.12}\n\nprices_lookup.keys() # dict_keys(['apple', 'oranges', 'milk'])\n\nprices_lookup.values() # dict_values([3.3, 3.56, 6.12])\n\nprices_lookup.items() # dict_items([('apple', 3.3), ('oranges', 3.56), ('milk', 6.12)])\n\n\n", "_____no_output_____" ], [ " d= {1:'david'}\n\n d", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code" ] ]

[ "FSFAP" ]

[ "FSFAP" ]

[ "FSFAP" ]

[ [ [ "<center>\n <img src=\"https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/IBMDeveloperSkillsNetwork-PY0101EN-SkillsNetwork/IDSNlogo.png\" width=\"300\" alt=\"cognitiveclass.ai logo\" />\n</center>\n\n# Classes and Objects in Python\n\nEstimated time needed: **40** minutes\n\n## Objectives\n\nAfter completing this lab you will be able to:\n\n- Work with classes and objects\n- Identify and define attributes and methods\n", "_____no_output_____" ], [ "<h2>Table of Contents</h2>\n<div class=\"alert alert-block alert-info\" style=\"margin-top: 20px\">\n <ul>\n <li>\n <a href=\"#intro\">Introduction to Classes and Objects</a>\n <ul>\n <li><a href=\"create\">Creating a class</a></li>\n <li><a href=\"instance\">Instances of a Class: Objects and Attributes</a></li>\n <li><a href=\"method\">Methods</a></li>\n </ul>\n </li>\n <li><a href=\"creating\">Creating a class</a></li>\n <li><a href=\"circle\">Creating an instance of a class Circle</a></li>\n <li><a href=\"rect\">The Rectangle Class</a></li>\n </ul>\n \n</div>\n\n<hr>\n", "_____no_output_____" ], [ "<h2 id=\"intro\">Introduction to Classes and Objects</h2>\n", "_____no_output_____" ], [ "<h3>Creating a Class</h3>\n", "_____no_output_____" ], [ "The first part of creating a class is giving it a name: In this notebook, we will create two classes, Circle and Rectangle. We need to determine all the data that make up that class, and we call that an attribute. Think about this step as creating a blue print that we will use to create objects. In figure 1 we see two classes, circle and rectangle. Each has their attributes, they are variables. The class circle has the attribute radius and color, while the rectangle has the attribute height and width. Let’s use the visual examples of these shapes before we get to the code, as this will help you get accustomed to the vocabulary.\n", "_____no_output_____" ], [ "<img src=\"https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/IBMDeveloperSkillsNetwork-PY0101EN-SkillsNetwork/labs/Module%203/images/ClassesClass.png\" width=\"500\" />\n", "_____no_output_____" ], [ "<i>Figure 1: Classes circle and rectangle, and each has their own attributes. The class circle has the attribute radius and colour, the rectangle has the attribute height and width.</i>\n", "_____no_output_____" ], [ "<h3 id=\"instance\">Instances of a Class: Objects and Attributes</h3>\n", "_____no_output_____" ], [ "An instance of an object is the realisation of a class, and in Figure 2 we see three instances of the class circle. We give each object a name: red circle, yellow circle and green circle. Each object has different attributes, so let's focus on the attribute of colour for each object.\n", "_____no_output_____" ], [ "<img src=\"https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/IBMDeveloperSkillsNetwork-PY0101EN-SkillsNetwork/labs/Module%203/images/ClassesObj.png\" width=\"500\" />\n", "_____no_output_____" ], [ "<i>Figure 2: Three instances of the class circle or three objects of type circle.</i>\n", "_____no_output_____" ], [ " The colour attribute for the red circle is the colour red, for the green circle object the colour attribute is green, and for the yellow circle the colour attribute is yellow. \n", "_____no_output_____" ], [ "<h3 id=\"method\">Methods</h3>\n", "_____no_output_____" ], [ "Methods give you a way to change or interact with the object; they are functions that interact with objects. For example, let’s say we would like to increase the radius by a specified amount of a circle. We can create a method called **add_radius(r)** that increases the radius by **r**. This is shown in figure 3, where after applying the method to the \"orange circle object\", the radius of the object increases accordingly. The “dot” notation means to apply the method to the object, which is essentially applying a function to the information in the object.\n", "_____no_output_____" ], [ "<img src=\"https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/IBMDeveloperSkillsNetwork-PY0101EN-SkillsNetwork/labs/Module%203/images/ClassesMethod.png\" width=\"500\" /> \n", "_____no_output_____" ], [ "<i>Figure 3: Applying the method “add_radius” to the object orange circle object.</i>\n", "_____no_output_____" ], [ "<hr>\n", "_____no_output_____" ], [ "<h2 id=\"creating\">Creating a Class</h2>\n", "_____no_output_____" ], [ "Now we are going to create a class circle, but first, we are going to import a library to draw the objects: \n", "_____no_output_____" ] ], [ [ "# Import the library\n\nimport matplotlib.pyplot as plt\n%matplotlib inline ", "_____no_output_____" ] ], [ [ " The first step in creating your own class is to use the <code>class</code> keyword, then the name of the class as shown in Figure 4. In this course the class parent will always be object: \n", "_____no_output_____" ], [ "<img src=\"https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/IBMDeveloperSkillsNetwork-PY0101EN-SkillsNetwork/labs/Module%203/images/ClassesDefine.png\" width=\"400\" />\n", "_____no_output_____" ], [ "<i>Figure 4: Creating a class Circle.</i>\n", "_____no_output_____" ], [ "The next step is a special method called a constructor <code>__init__</code>, which is used to initialize the object. The input are data attributes. The term <code>self</code> contains all the attributes in the set. For example the <code>self.color</code> gives the value of the attribute color and <code>self.radius</code> will give you the radius of the object. We also have the method <code>add_radius()</code> with the parameter <code>r</code>, the method adds the value of <code>r</code> to the attribute radius. To access the radius we use the syntax <code>self.radius</code>. The labeled syntax is summarized in Figure 5:\n", "_____no_output_____" ], [ "<img src=\"https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/IBMDeveloperSkillsNetwork-PY0101EN-SkillsNetwork/labs/Module%203/images/ClassesCircle.png\" width=\"600\" />\n", "_____no_output_____" ], [ "<i>Figure 5: Labeled syntax of the object circle.</i>\n", "_____no_output_____" ], [ "The actual object is shown below. We include the method <code>drawCircle</code> to display the image of a circle. We set the default radius to 3 and the default colour to blue:\n", "_____no_output_____" ] ], [ [ "# Create a class Circle\n\nclass Circle(object):\n \n # Constructor\n def __init__(self, radius=3, color='blue'):\n self.radius = radius\n self.color = color \n \n # Method\n def add_radius(self, r):\n self.radius = self.radius + r\n return(self.radius)\n \n # Method\n def drawCircle(self):\n plt.gca().add_patch(plt.Circle((0, 0), radius=self.radius, fc=self.color))\n plt.axis('scaled')\n plt.show() ", "_____no_output_____" ] ], [ [ "<hr>\n", "_____no_output_____" ], [ "<h2 id=\"circle\">Creating an instance of a class Circle</h2>\n", "_____no_output_____" ], [ "Let’s create the object <code>RedCircle</code> of type Circle to do the following:\n", "_____no_output_____" ] ], [ [ "# Create an object RedCircle\n\nRedCircle = Circle(10, 'red')", "_____no_output_____" ] ], [ [ "We can use the <code>dir</code> command to get a list of the object's methods. Many of them are default Python methods.\n", "_____no_output_____" ] ], [ [ "# Find out the methods can be used on the object RedCircle\n\ndir(RedCircle)", "_____no_output_____" ] ], [ [ "We can look at the data attributes of the object: \n", "_____no_output_____" ] ], [ [ "# Print the object attribute radius\n\nRedCircle.radius", "_____no_output_____" ], [ "# Print the object attribute color\n\nRedCircle.color", "_____no_output_____" ] ], [ [ " We can change the object's data attributes: \n", "_____no_output_____" ] ], [ [ "# Set the object attribute radius\n\nRedCircle.radius = 1\nRedCircle.radius", "_____no_output_____" ] ], [ [ " We can draw the object by using the method <code>drawCircle()</code>:\n", "_____no_output_____" ] ], [ [ "# Call the method drawCircle\n\nRedCircle.drawCircle()", "_____no_output_____" ] ], [ [ "We can increase the radius of the circle by applying the method <code>add_radius()</code>. Let increases the radius by 2 and then by 5: \n", "_____no_output_____" ] ], [ [ "# Use method to change the object attribute radius\n\nprint('Radius of object:',RedCircle.radius)\nRedCircle.add_radius(2)\nprint('Radius of object of after applying the method add_radius(2):',RedCircle.radius)\nRedCircle.add_radius(5)\nprint('Radius of object of after applying the method add_radius(5):',RedCircle.radius)\nRedCircle.add_radius(6)\nprint('Radius of object of after applying the method add radius(6):',RedCircle.radius)", "Radius of object: 1\nRadius of object of after applying the method add_radius(2): 3\nRadius of object of after applying the method add_radius(5): 8\nRadius of object of after applying the method add radius(6): 14\n" ] ], [ [ " Let’s create a blue circle. As the default colour is blue, all we have to do is specify what the radius is:\n", "_____no_output_____" ] ], [ [ "# Create a blue circle with a given radius\n\nBlueCircle = Circle(radius=100)", "_____no_output_____" ] ], [ [ " As before we can access the attributes of the instance of the class by using the dot notation:\n", "_____no_output_____" ] ], [ [ "# Print the object attribute radius\n\nBlueCircle.radius", "_____no_output_____" ], [ "# Print the object attribute color\n\nBlueCircle.color", "_____no_output_____" ] ], [ [ " We can draw the object by using the method <code>drawCircle()</code>:\n", "_____no_output_____" ] ], [ [ "# Call the method drawCircle\n\nBlueCircle.drawCircle()", "_____no_output_____" ] ], [ [ "Compare the x and y axis of the figure to the figure for <code>RedCircle</code>; they are different.\n", "_____no_output_____" ], [ "<hr>\n", "_____no_output_____" ], [ "<h2 id=\"rect\">The Rectangle Class</h2>\n", "_____no_output_____" ], [ "Let's create a class rectangle with the attributes of height, width and color. We will only add the method to draw the rectangle object:\n", "_____no_output_____" ] ], [ [ "# Create a new Rectangle class for creating a rectangle object\n\nclass Rectangle(object):\n \n # Constructor\n def __init__(self, width=2, height=3, color='r'):\n self.height = height \n self.width = width\n self.color = color\n \n # Method\n def drawRectangle(self):\n plt.gca().add_patch(plt.Rectangle((0, 0), self.width, self.height ,fc=self.color))\n plt.axis('scaled')\n plt.show()\n ", "_____no_output_____" ] ], [ [ "Let’s create the object <code>SkinnyBlueRectangle</code> of type Rectangle. Its width will be 2 and height will be 3, and the color will be blue:\n", "_____no_output_____" ] ], [ [ "# Create a new object rectangle\n\nSkinnyBlueRectangle = Rectangle(2, 10, 'blue')", "_____no_output_____" ] ], [ [ " As before we can access the attributes of the instance of the class by using the dot notation:\n", "_____no_output_____" ] ], [ [ "# Print the object attribute height\n\nSkinnyBlueRectangle.height ", "_____no_output_____" ], [ "# Print the object attribute width\n\nSkinnyBlueRectangle.width", "_____no_output_____" ], [ "# Print the object attribute color\n\nSkinnyBlueRectangle.color", "_____no_output_____" ] ], [ [ " We can draw the object:\n", "_____no_output_____" ] ], [ [ "# Use the drawRectangle method to draw the shape\n\nSkinnyBlueRectangle.drawRectangle()", "_____no_output_____" ] ], [ [ "Let’s create the object <code>FatYellowRectangle</code> of type Rectangle :\n", "_____no_output_____" ] ], [ [ "# Create a new object rectangle\n\nFatYellowRectangle = Rectangle(20, 5, 'yellow')", "_____no_output_____" ] ], [ [ " We can access the attributes of the instance of the class by using the dot notation:\n", "_____no_output_____" ] ], [ [ "# Print the object attribute height\n\nFatYellowRectangle.height ", "_____no_output_____" ], [ "# Print the object attribute width\n\nFatYellowRectangle.width", "_____no_output_____" ], [ "# Print the object attribute color\n\nFatYellowRectangle.color", "_____no_output_____" ] ], [ [ " We can draw the object:\n", "_____no_output_____" ] ], [ [ "# Use the drawRectangle method to draw the shape\n\nFatYellowRectangle.drawRectangle()", "_____no_output_____" ] ], [ [ "<hr>\n", "_____no_output_____" ], [ "<h2 id=\"rect\">Exercises</h2>\n", "_____no_output_____" ], [ "<h4> Text Analysis </h4>\n", "_____no_output_____" ], [ "You have been recruited by your friend, a linguistics enthusiast, to create a utility tool that can perform analysis on a given piece of text. Complete the class\n'analysedText' with the following methods -\n\n<ul>\n <li> Constructor - Takes argument 'text',makes it lower case and removes all punctuation. Assume only the following punctuation is used - period (.), exclamation mark (!), comma (,) and question mark (?). Store the argument in \"fmtText\" \n <li> freqAll - returns a dictionary of all unique words in the text along with the number of their occurences.\n <li> freqOf - returns the frequency of the word passed in argument.\n</ul>\n The skeleton code has been given to you. Docstrings can be ignored for the purpose of the exercise. <br>\n <i> Hint: Some useful functions are <code>replace()</code>, <code>lower()</code>, <code>split()</code>, <code>count()</code> </i><br>\n", "_____no_output_____" ] ], [ [ "class analysedText(object):\n \n def __init__ (self, text):\n \n reArrText = text.lower()\n \n reArrText = reArrText.replace('.','').replace('!','').replace(',','').replace('?','')\n \n self.fmtText = reArrText\n \n def freqAll(self): \n \n wordList = self.fmtText.split(' ')\n \n \n freqMap = {}\n for word in set(wordList): # use set to remove duplicates in list\n freqMap[word] = wordList.count(word)\n \n return freqMap\n \n def freqOf(self,word):\n \n freqDict = self.freqAll()\n \n if word in freqDict:\n return freqDict[word]\n else:\n return 0\n", "_____no_output_____" ] ], [ [ "Execute the block below to check your progress.\n", "_____no_output_____" ] ], [ [ "import sys\n\nsampleMap = {'eirmod': 1,'sed': 1, 'amet': 2, 'diam': 5, 'consetetur': 1, 'labore': 1, 'tempor': 1, 'dolor': 1, 'magna': 2, 'et': 3, 'nonumy': 1, 'ipsum': 1, 'lorem': 2}\n\ndef testMsg(passed):\n if passed:\n return 'Test Passed'\n else :\n return 'Test Failed'\n\nprint(\"Constructor: \")\ntry:\n samplePassage = analysedText(\"Lorem ipsum dolor! diam amet, consetetur Lorem magna. sed diam nonumy eirmod tempor. diam et labore? et diam magna. et diam amet.\")\n print(testMsg(samplePassage.fmtText == \"lorem ipsum dolor diam amet consetetur lorem magna sed diam nonumy eirmod tempor diam et labore et diam magna et diam amet\"))\nexcept:\n print(\"Error detected. Recheck your function \" )\nprint(\"freqAll: \",)\ntry:\n wordMap = samplePassage.freqAll()\n print(testMsg(wordMap==sampleMap))\nexcept:\n print(\"Error detected. Recheck your function \" )\nprint(\"freqOf: \")\ntry:\n passed = True\n for word in sampleMap:\n if samplePassage.freqOf(word) != sampleMap[word]:\n passed = False\n break\n print(testMsg(passed))\n \nexcept:\n print(\"Error detected. Recheck your function \" )\n ", "Constructor: \nTest Passed\nfreqAll: \nTest Passed\nfreqOf: \nTest Passed\n" ] ], [ [ "<details><summary>Click here for the solution</summary>\n\n```python\nclass analysedText(object):\n \n def __init__ (self, text):\n # remove punctuation\n formattedText = text.replace('.','').replace('!','').replace('?','').replace(',','')\n \n # make text lowercase\n formattedText = formattedText.lower()\n \n self.fmtText = formattedText\n \n def freqAll(self): \n # split text into words\n wordList = self.fmtText.split(' ')\n \n # Create dictionary\n freqMap = {}\n for word in set(wordList): # use set to remove duplicates in list\n freqMap[word] = wordList.count(word)\n \n return freqMap\n \n def freqOf(self,word):\n # get frequency map\n freqDict = self.freqAll()\n \n if word in freqDict:\n return freqDict[word]\n else:\n return 0\n \n```\n\n</details>\n \n", "_____no_output_____" ], [ "<hr>\n<h2>The last exercise!</h2>\n<p>Congratulations, you have completed your first lesson and hands-on lab in Python. However, there is one more thing you need to do. The Data Science community encourages sharing work. The best way to share and showcase your work is to share it on GitHub. By sharing your notebook on GitHub you are not only building your reputation with fellow data scientists, but you can also show it off when applying for a job. Even though this was your first piece of work, it is never too early to start building good habits. So, please read and follow <a href=\"https://cognitiveclass.ai/blog/data-scientists-stand-out-by-sharing-your-notebooks/\" target=\"_blank\">this article</a> to learn how to share your work.\n<hr>\n", "_____no_output_____" ], [ "## Author\n\n<a href=\"https://www.linkedin.com/in/joseph-s-50398b136/\" target=\"_blank\">Joseph Santarcangelo</a>\n\n## Other contributors\n\n<a href=\"www.linkedin.com/in/jiahui-mavis-zhou-a4537814a\">Mavis Zhou</a>\n\n## Change Log\n\n| Date (YYYY-MM-DD) | Version | Changed By | Change Description |\n| ----------------- | ------- | ---------- | ---------------------------------- |\n| 2020-08-26 | 2.0 | Lavanya | Moved lab to course repo in GitLab |\n| | | | |\n| | | | |\n\n<hr/>\n\n## <h3 align=\"center\"> © IBM Corporation 2020. All rights reserved. <h3/>\n", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Analyzing the Effects of Non-Academic Features on Student Performance", "_____no_output_____" ] ], [ [ "# For reading data sets\nimport pandas\n# For lots of awesome things\nimport numpy as np\n# Need this for LabelEncoder\nfrom sklearn import preprocessing\n# For building our net\nimport keras\n# For plotting\nimport matplotlib.pyplot as plt\n%matplotlib inline", "_____no_output_____" ] ], [ [ "## Read in data", "_____no_output_____" ], [ "### Data is seperated by a semicolon (delimiter=\";\") containing column names as the first row of the file (header = 0).", "_____no_output_____" ] ], [ [ "# Read in student data\nstudent_data = np.array(pandas.read_table(\"./student-por.csv\",\ndelimiter=\";\", header=0))\n\n# Display student data\nstudent_data", "_____no_output_____" ] ], [ [ "### Determine what the column labels are...", "_____no_output_____" ] ], [ [ "# Descriptions for each feature (found in the header)\nfeature_descrips = np.array(pandas.read_csv(\"./student-por.csv\",\ndelimiter=\";\", header=None, nrows=1))\n\n# Display descriptions\nprint(feature_descrips)", "[['school' 'sex' 'age' 'address' 'famsize' 'Pstatus' 'Medu' 'Fedu' 'Mjob'\n 'Fjob' 'reason' 'guardian' 'traveltime' 'studytime' 'failures'\n 'schoolsup' 'famsup' 'paid' 'activities' 'nursery' 'higher' 'internet'\n 'romantic' 'famrel' 'freetime' 'goout' 'Dalc' 'Walc' 'health'\n 'absences' 'G1' 'G2' 'G3']]\n" ] ], [ [ "### ...and give them clearer descriptions.", "_____no_output_____" ] ], [ [ "# More detailed descriptions\nfeature_descrips = np.array([\"School\", \"Sex\", \"Age\", \"Urban or Rural Address\", \"Family Size\", \n \"Parent's Cohabitation status\", \"Mother's Education\", \"Father's Education\",\n \"Mother's Job\", \"Father's Job\", \"Reason for Choosing School\", \n \"Student's Gaurdain\", \"Home to School Travel Time\", \"Weekly Study Time\",\n \"Number of Past Class Failures\", \"Extra Educational Support\", \n \"Family Educational Support\", \"Extra Paid Classes\", \"Extra Curricular Activities\",\n \"Attended Nursery School\", \"Wants to Take Higher Education\", \"Internet Access at Home\", \n \"In a Romantic Relationship\", \"Quality of Family Relationships\",\n \"Free Time After School\", \"Time Spent Going out With Friends\", \n \"Workday Alcohol Consumption\", \"Weekend Alcohol Consumption\", \n \"Current Health Status\", \"Number of Student Absences\", \"First Period Grade\",\n \"Second Period Grade\", \"Final Grade\"])", "_____no_output_____" ] ], [ [ "# Data Cleanup", "_____no_output_____" ], [ "## Shuffle data", "_____no_output_____" ], [ "### We sampled 2 schools, and right now our data has each school grouped together. We need to get rid of this grouping for training later down the road.", "_____no_output_____" ] ], [ [ "# Shuffle the data!\nnp.random.shuffle(student_data)\nstudent_data", "_____no_output_____" ] ], [ [ "## Alphabetically classify scores", "_____no_output_____" ], [ "### Because our data is sampled from Portugal, we have to modify their scoring system a bit to represent something more like ours.\n### 0 = F\n### 1 = D\n### 2 = C\n### 3 = B\n### 4 = A", "_____no_output_____" ] ], [ [ "# Array holding final scores for every student\nscores = student_data[:,32]\n\n# Iterate through list of scores, changing them from a 0-19 value\n## to a 0-4 value (representing F-A)\nfor i in range(len(scores)):\n if(scores[i] > 18):\n scores[i] = 4\n elif(scores[i] > 16):\n scores[i] = 3\n elif(scores[i] > 14):\n scores[i] = 2\n elif(scores[i] > 12):\n scores[i] = 1\n else:\n scores[i] = 0\n \n# Update the final scores in student_data to reflect these changes\nfor i in range(len(scores)):\n student_data[i,32] = scores[i]\n \n# Display new data. Hint: Look at the last column\nstudent_data", "_____no_output_____" ] ], [ [ "## Encoding non-numeric data to integers", "_____no_output_____" ] ], [ [ "# One student sample\nstudent_data[0,:]", "_____no_output_____" ] ], [ [ "### We have some qualitative data from the questionaire that needs to be converted to represent numbers.", "_____no_output_____" ] ], [ [ "# Label Encoder\nle = preprocessing.LabelEncoder()\n\n# Columns that hold non-numeric data\nindices = np.array([0,1,3,4,5,8,9,10,11,15,16,17,18,19,20,21,22])\n\n# Transform the non-numeric data in these columns to integers\nfor i in range(len(indices)):\n column = indices[i]\n le.fit(student_data[:,column])\n student_data[:,column] = le.transform(student_data[:,column])", "_____no_output_____" ], [ "student_data[0,:]", "_____no_output_____" ] ], [ [ "## Encoding 0's to -1 for binomial data.", "_____no_output_____" ], [ "### We want our weights to change because 0 represents something! Therefore, we need to encode 0's to -1's so the weights will change with that input.", "_____no_output_____" ] ], [ [ "# Columns that hold binomial data\nindices = np.array([0,1,3,4,5,15,16,17,18,19,20,21,22])\n\n# Change 0's to -1's\nfor i in range(len(indices)):\n column = indices[i]\n \n # values of current feature\n feature = student_data[:,column]\n \n # change values to -1 if equal to 0\n feature = np.where(feature==0, -1, feature)\n student_data[:,column] = feature\n \nstudent_data[0,:]", "_____no_output_____" ] ], [ [ "## Standardizing the nominal and numerical data.", "_____no_output_____" ], [ "### We need our input to matter equally (Everyone is important!). We do this by standardizing our data (get a mean of 0 and a stardard deviation of 1).", "_____no_output_____" ] ], [ [ "scaler = preprocessing.StandardScaler()", "_____no_output_____" ], [ "temp = student_data[:,[2,6,7,8,9,10,11,12,13,14,23,24,25,26,27,28,29,30,31]]\nprint(student_data[0,:])\nStandardized = scaler.fit_transform(temp)", "[-1 -1 16 1 -1 1 4 3 1 2 1 1 1 2 0 -1 1 -1 1 1 1 1 -1 4 3 5 1 5 2 2 14 14\n 2]\n" ], [ "print('Mean:', round(Standardized.mean()))\nprint('Standard deviation:', Standardized.std())", "Mean: -0.0\nStandard deviation: 1.0\n" ], [ "student_data[:,[2,6,7,8,9,10,11,12,13,14,23,24,25,26,27,28,29,30,31]] = Standardized", "_____no_output_____" ], [ "student_data[0,:]", "_____no_output_____" ] ], [ [ "## Convert results to one-hot encoding", "_____no_output_____" ] ], [ [ "# Final grades\nresults = student_data[:,32]\n\n# Take a look at first 5 final grades\nprint(\"First 5 final grades:\", results[0:5])\n\n# All unique values for final grades (0-4 representing F-A)\npossible_results = np.unique(student_data[:,32]).T\nprint(\"All possible results:\", possible_results)", "First 5 final grades: [2 3 0 1 1]\nAll possible results: [0 1 2 3 4]\n" ], [ "# One-hot encode final grades (results) which will be used as our output\n# The length of the \"ID\" should be as long as the total number of possible results so each results\n## gets its own, personal one-hot encoding\ny = keras.utils.to_categorical(results,len(possible_results))\n\n# Take a look at the first 5 final grades now (no longer numbers but arrays)\ny[0:5]", "_____no_output_____" ], [ "# our input, all features except final grades\nx = student_data[:,0:32]", "_____no_output_____" ] ], [ [ "# Model Building", "_____no_output_____" ], [ "#### Now let's create a function that will build a model for us. This will come in handy later on. Our model will have two hidden layers. The first hidden layer will have an input size of 800, and the second will have an input size of 400. The optimizer that we are using is adamax which is good at ignoring noise in a datset. The loss function we are using is called categorical cross entropy and which is useful for trying to classify or label something. In this case, we are trying to classify students by letter grades, so this loss function will be of great use to us.", "_____no_output_____" ] ], [ [ "# Function to create network given model\ndef create_network(model):\n # Specify input/output size\n input_size = x.shape[1]\n output_size = y.shape[1]\n\n # Create the hidden layer\n model.add(keras.layers.Dense(800, input_dim = input_size, activation = 'relu'))\n\n # Additional hidden layer\n model.add(keras.layers.Dense(400,activation='relu'))\n\n # Output layer\n model.add(keras.layers.Dense(output_size,activation='softmax'))\n\n # Compile - why using adamax?\n model.compile(loss='categorical_crossentropy',\n optimizer='adamax', \n metrics=['accuracy'])", "_____no_output_____" ], [ "# Feed-forward model\nmodel = keras.Sequential()\ncreate_network(model)", "_____no_output_____" ] ], [ [ "# Initial Test of the Network", "_____no_output_____" ] ], [ [ "# Split data into training and testing data\nx_train = x[0:518,:]\nx_test = x[519:649,:]\n\ny_train = y[0:518,:]\ny_test = y[519:649,:]", "_____no_output_____" ], [ "# Train on training data! \n# We're saving this information in the variable -history- so we can take a look at it later\nhistory = model.fit(x_train, y_train,\n batch_size = 32, \n epochs = 7, \n verbose = 0, \n validation_split = 0.2)", "_____no_output_____" ], [ "# Validate using data the network hasn't seen before (testing data)\n# Save this info in -score- so we can take a look at it\nscore = model.evaluate(x_test,y_test, verbose=0)\n\n# Check it's effectiveness\nprint('Test loss:', score[0])\nprint('Test accuracy:', score[1])", "Test loss: 0.49870656898173576\nTest accuracy: 0.7692307692307693\n" ], [ "# Plot the data\ndef plot(history):\n\n plt.figure(1)\n\n # Summarize history for accuracy\n\n plt.subplot(211)\n plt.plot(history.history['acc'])\n plt.plot(history.history['val_acc'])\n plt.title('model accuracy')\n plt.ylabel('accuracy')\n plt.xlabel('epoch')\n plt.legend(['train','test'], loc ='upper left')\n\n # Summarize history for loss\n\n plt.subplot(212)\n plt.plot(history.history['loss'])\n plt.plot(history.history['val_loss'])\n plt.title('model loss')\n plt.ylabel('loss')\n plt.xlabel('epoch')\n plt.legend(['train','test'], loc ='upper left')\n\n # Display plot\n \n plt.tight_layout()\n plt.show()", "_____no_output_____" ], [ "# Plot current training and validation accuracy and loss\nplot(history)", "_____no_output_____" ] ], [ [ "# Training and Testing Without Individual Features", "_____no_output_____" ] ], [ [ "# Analyze the effects of removing one feature on training\ndef remove_and_analyze(feature):\n # Told you those feature descriptions would be useful\n print(\"Without feature\", feature, \":\", feature_descrips[feature])\n \n # Create feed-forward network\n model = keras.Sequential()\n create_network(model)\n \n # Remove feature from columns (axis 1)\n x = np.delete(student_data, feature, axis = 1)\n \n # Split data into training and testing data\n x_train = x[0:518,:]\n x_test = x[519:649,:]\n \n # Train on training data!\n history = model.fit(x_train, y_train,\n batch_size = 32, \n epochs = 7, \n verbose = 0, \n validation_split = 0.2)\n \n # Validate using data the network hasn't seen before (testing data)\n score = model.evaluate(x_test,y_test, verbose=0)\n \n # Check it's effectiveness\n print('Test loss:', score[0])\n print('Test accuracy:', score[1])\n \n # Plot the data\n plot(history)", "_____no_output_____" ], [ "# Analyze the effects of removing one feature on training\n# Do this for all input features\nfor i in range(student_data.shape[1]-1):\n remove_and_analyze(i)\n print(\"\\n \\n \\n\")", "Without feature 0 : School\nTest loss: 0.1621086014179477\nTest accuracy: 0.9461538461538461\n" ] ], [ [ "# Training and Testing Without Five Features", "_____no_output_____" ] ], [ [ "# Delete the five features that most negatively impact accuracy\nx = np.delete(student_data, 21, axis = 1)\nx = np.delete(x, 20, axis = 1)\nx = np.delete(x, 9, axis = 1)\nx = np.delete(x, 8, axis = 1)\nx = np.delete(x, 7, axis = 1)\n\n# Create feed-forward network\nmodel = keras.Sequential()\ncreate_network(model)\n\n# Split data into training and testing data\nx_train = x[0:518,:]\nx_test = x[519:649,:]\n\n# Train on training data!\nhistory = model.fit(x_train, y_train,\n batch_size = 32, \n epochs = 7, \n verbose = 0, \n validation_split = 0.2)\n\n# Validate using data the network hasn't seen before (testing data)\nscore = model.evaluate(x_test,y_test, verbose=0)\n\n# Check it's effectiveness\nprint('Test loss:', score[0])\nprint('Test accuracy:', score[1])\n\n# Plot the data\nplot(history)", "Test loss: 0.1731401116976765\nTest accuracy: 0.9307692307692308\n" ] ], [ [ "# Grade Distribution Analysis", "_____no_output_____" ] ], [ [ "# Function for analyzing the percent of students with each grade [F,D,C,B,A]\ndef analyze(array):\n \n # To hold the total number of students with a certain final grade\n # Index 0 - F. Index 4 - A\n sums = np.array([0,0,0,0,0])\n \n # Iterate through array. Update sums according to whether a student got a final grade of a(n)\n for i in range(len(array)):\n # F\n if(array[i]==0):\n sums[0] += 1\n # D\n elif(array[i]==1):\n sums[1] +=1\n # C\n elif(array[i]==2):\n sums[2] +=1\n # B\n elif(array[i]==3):\n sums[3] +=1\n # A\n else:\n sums[4] += 1\n \n # Total number of students\n total = sums[0] + sums[1] + sums[2] + sums[3] + sums[4]\n \n # Hold percentage of students with grade of [F,D,C,B,A]\n percentages = np.array([sums[0]/total*100, \n sums[1]/total*100, \n sums[2]/total*100, \n sums[3]/total*100, \n sums[4]/total*100])\n \n # One bar for each of the 5 grades\n x = np.array([1,2,3,4,5])\n \n # Descriptions for each bar. None on y-axis\n plt.xticks(np.arange(6), ('', 'F', 'D', 'C', 'B','A'))\n \n # X axis - grades. Y axis - percentage of students with each grade\n plt.bar(x,percentages)\n plt.xlabel(\"Grades\")\n plt.ylabel(\"Percentage of Students\")\n \n # Display bar graph\n plt.show()\n \n # Display percentages\n print(percentages)", "_____no_output_____" ] ], [ [ "## Family Educational Support", "_____no_output_____" ] ], [ [ "# Array holding final grades of all students who have family educational support\nfam_sup = []\n# Array holding final grades of all students who have family educational support\nno_fam_sup = []\n\n# Iterate through all student samples\nfor i in range(student_data.shape[0]):\n \n # Does the student have family educational support? (-1 no, 1 yes)\n sup = student_data[i][16]\n \n # Append student's final grade to corresponding array\n if(sup==1):\n fam_sup.append(student_data[i][32])\n else:\n no_fam_sup.append(student_data[i][32])", "_____no_output_____" ] ], [ [ "### Family Educational Support", "_____no_output_____" ] ], [ [ "analyze(fam_sup)", "_____no_output_____" ] ], [ [ "### No Family Educational Support", "_____no_output_____" ] ], [ [ "analyze(no_fam_sup)", "_____no_output_____" ] ], [ [ "## Reason for choosing school", "_____no_output_____" ] ], [ [ "# Each array holds the grades of students who chose to go to their school for that reason\n# Close to home\nreason1 = []\n# School reputation\nreason2 = []\n# Course prefrence\nreason3 = []\n# Other\nreason4 = []\n\n# Values that represent these unique reasons. They are not integer numbers like in the previous\n## example. They're floatig point numbers so we'll save them so we can compare them to the value\n## of this feature in each sample\nunique_reasons = np.unique(student_data[:,10])\n\n# Iterate through all student samples and append final grades to corresponding arrays \nfor i in range(student_data.shape[0]):\n \n reason = student_data[i][10]\n \n if(reason==unique_reasons[0]):\n reason1.append(student_data[i][32])\n elif(reason==unique_reasons[1]):\n reason2.append(student_data[i][32])\n elif(reason==unique_reasons[2]):\n reason3.append(student_data[i][32])\n else:\n reason4.append(student_data[i][32])", "_____no_output_____" ] ], [ [ "### Reason 1: Close to Home", "_____no_output_____" ] ], [ [ "analyze(reason1)", "_____no_output_____" ] ], [ [ "### Reason 2: School Reputation", "_____no_output_____" ] ], [ [ "analyze(reason2)", "_____no_output_____" ] ], [ [ "### Reason 3: Course Prefrence", "_____no_output_____" ] ], [ [ "analyze(reason3)", "_____no_output_____" ] ], [ [ "### Reason 4: Other", "_____no_output_____" ] ], [ [ "analyze(reason4)", "_____no_output_____" ] ], [ [ "## Frequency of Going Out With Friends", "_____no_output_____" ] ], [ [ "# Each array holds the grades of students who go out with friends for that specified amount of time \n# (1 - very low, 5 - very high)\ngo_out1 = []\ngo_out2 = []\ngo_out3 = []\ngo_out4 = []\ngo_out5 = []\n\n# Floating point values representing frequency\nunique = np.unique(student_data[:,25])\n\n# Iterate through all student samples and append final grades to corresponding arrays \nfor i in range(student_data.shape[0]):\n \n frequency = student_data[i][25]\n \n if(frequency==unique[0]):\n go_out1.append(student_data[i][32])\n elif(frequency==unique[1]):\n go_out2.append(student_data[i][32])\n elif(frequency==unique[2]):\n go_out3.append(student_data[i][32])\n elif(frequency==unique[3]):\n go_out4.append(student_data[i][32])\n else:\n go_out5.append(student_data[i][32])", "_____no_output_____" ], [ "analyze(go_out1)\nanalyze(go_out2)\nanalyze(go_out3)\nanalyze(go_out4)\nanalyze(go_out5)", "_____no_output_____" ] ], [ [ "## Free Time after School", "_____no_output_____" ] ], [ [ "# Each array holds the grades of students who have the specified amount of free time after school \n# (1 - very low, 5 - very high)\nfree1 = []\nfree2 = []\nfree3 = []\nfree4 = []\nfree5 = []\n\n# Floating point values representing frequency\nunique = np.unique(student_data[:,24])\n\n# Iterate through all student samples and append final grades to corresponding arrays \nfor i in range(student_data.shape[0]):\n \n frequency = student_data[i][24]\n \n if(frequency==unique[0]):\n free1.append(student_data[i][32])\n elif(frequency==unique[1]):\n free2.append(student_data[i][32])\n elif(frequency==unique[2]):\n free3.append(student_data[i][32])\n elif(frequency==unique[3]):\n free4.append(student_data[i][32])\n else:\n free5.append(student_data[i][32])", "_____no_output_____" ], [ "analyze(free1)\nanalyze(free2)\nanalyze(free3)\nanalyze(free4)\nanalyze(free5)", "_____no_output_____" ] ], [ [ "## Paid Classes", "_____no_output_____" ] ], [ [ "# Array holding final grades of all students who have extra paid classes\npaid_class = []\n# Array holding final grades of all students who do not have extra paid classes\nno_paid_class = []\n\n# Iterate through all student samples and append final grades to corresponding arrays \nfor i in range(student_data.shape[0]):\n \n paid = student_data[i][17]\n \n if(paid==1):\n paid_class.append(student_data[i][32])\n else:\n no_paid_class.append(student_data[i][32])", "_____no_output_____" ] ], [ [ "### Extra Paid Classes", "_____no_output_____" ] ], [ [ "analyze(paid_class)", "_____no_output_____" ] ], [ [ "### No Extra Paid Classes", "_____no_output_____" ] ], [ [ "analyze(no_paid_class)", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "Unlicense" ]

[ "Unlicense" ]

[ "Unlicense" ]

[ [ [ "users = [\n{ \"id\": 0, \"name\": \"Hero\" },\n{ \"id\": 1, \"name\": \"Dunn\" },\n{ \"id\": 2, \"name\": \"Sue\" },\n{ \"id\": 3, \"name\": \"Chi\" },\n{ \"id\": 4, \"name\": \"Thor\" },\n{ \"id\": 5, \"name\": \"Clive\" },\n{ \"id\": 6, \"name\": \"Hicks\" },\n{ \"id\": 7, \"name\": \"Devin\" },\n{ \"id\": 8, \"name\": \"Kate\" },\n{ \"id\": 9, \"name\": \"Klein\" }\n]\n\n# “friendship” data, represented as a list of pairs of IDs\n# the tuple (0, 1) indicates that the data scientist with id 0 (Hero) and\n# the data scientist with id 1 (Dunn) are friends.\n\nfriendships = [(0, 1), (0, 2), (1, 2), (1, 3), (2, 3), (3, 4),\n (4, 5), (5, 6), (5, 7), (6, 8), (7, 8), (8, 9)]", "_____no_output_____" ] ], [ [ "# 1) Add a list of friends to each user", "_____no_output_____" ] ], [ [ "# set each user’s friends property to an empty list:\nfor user in users:\n user[\"friends\"] = []\n\nprint users", "[{'friends': [], 'id': 0, 'name': 'Hero'}, {'friends': [], 'id': 1, 'name': 'Dunn'}, {'friends': [], 'id': 2, 'name': 'Sue'}, {'friends': [], 'id': 3, 'name': 'Chi'}, {'friends': [], 'id': 4, 'name': 'Thor'}, {'friends': [], 'id': 5, 'name': 'Clive'}, {'friends': [], 'id': 6, 'name': 'Hicks'}, {'friends': [], 'id': 7, 'name': 'Devin'}, {'friends': [], 'id': 8, 'name': 'Kate'}, {'friends': [], 'id': 9, 'name': 'Klein'}]\n" ], [ "print users[0]['friends']", "[]\n" ], [ "# then we populate the lists using the friendships data:\n\nfor i, j in friendships:\n # this works because users[i] is the user whose id is i\n users[i][\"friends\"].append(users[j]) # add i as a friend of j\n users[j][\"friends\"].append(users[i]) # add j as a friend of i\nprint users[0]", "{'friends': [{'friends': [{...}, {'friends': [{...}, {...}, {'friends': [{...}, {...}, {'friends': [{...}, {'friends': [{...}, {'friends': [{...}, {'friends': [{...}, {'friends': [{...}, {...}], 'id': 7, 'name': 'Devin'}, {'friends': [{...}], 'id': 9, 'name': 'Klein'}], 'id': 8, 'name': 'Kate'}], 'id': 6, 'name': 'Hicks'}, {'friends': [{...}, {'friends': [{'friends': [{...}, {...}], 'id': 6, 'name': 'Hicks'}, {...}, {'friends': [{...}], 'id': 9, 'name': 'Klein'}], 'id': 8, 'name': 'Kate'}], 'id': 7, 'name': 'Devin'}], 'id': 5, 'name': 'Clive'}], 'id': 4, 'name': 'Thor'}], 'id': 3, 'name': 'Chi'}], 'id': 2, 'name': 'Sue'}, {'friends': [{...}, {'friends': [{...}, {...}, {...}], 'id': 2, 'name': 'Sue'}, {'friends': [{...}, {'friends': [{...}, {'friends': [{...}, {'friends': [{...}, {'friends': [{...}, {...}], 'id': 7, 'name': 'Devin'}, {'friends': [{...}], 'id': 9, 'name': 'Klein'}], 'id': 8, 'name': 'Kate'}], 'id': 6, 'name': 'Hicks'}, {'friends': [{...}, {'friends': [{'friends': [{...}, {...}], 'id': 6, 'name': 'Hicks'}, {...}, {'friends': [{...}], 'id': 9, 'name': 'Klein'}], 'id': 8, 'name': 'Kate'}], 'id': 7, 'name': 'Devin'}], 'id': 5, 'name': 'Clive'}], 'id': 4, 'name': 'Thor'}], 'id': 3, 'name': 'Chi'}], 'id': 1, 'name': 'Dunn'}, {'friends': [{...}, {'friends': [{...}, {...}, {'friends': [{...}, {...}, {'friends': [{...}, {'friends': [{...}, {'friends': [{...}, {'friends': [{...}, {'friends': [{...}, {...}], 'id': 7, 'name': 'Devin'}, {'friends': [{...}], 'id': 9, 'name': 'Klein'}], 'id': 8, 'name': 'Kate'}], 'id': 6, 'name': 'Hicks'}, {'friends': [{...}, {'friends': [{'friends': [{...}, {...}], 'id': 6, 'name': 'Hicks'}, {...}, {'friends': [{...}], 'id': 9, 'name': 'Klein'}], 'id': 8, 'name': 'Kate'}], 'id': 7, 'name': 'Devin'}], 'id': 5, 'name': 'Clive'}], 'id': 4, 'name': 'Thor'}], 'id': 3, 'name': 'Chi'}], 'id': 1, 'name': 'Dunn'}, {'friends': [{'friends': [{...}, {...}, {...}], 'id': 1, 'name': 'Dunn'}, {...}, {'friends': [{...}, {'friends': [{...}, {'friends': [{...}, {'friends': [{...}, {'friends': [{...}, {...}], 'id': 7, 'name': 'Devin'}, {'friends': [{...}], 'id': 9, 'name': 'Klein'}], 'id': 8, 'name': 'Kate'}], 'id': 6, 'name': 'Hicks'}, {'friends': [{...}, {'friends': [{'friends': [{...}, {...}], 'id': 6, 'name': 'Hicks'}, {...}, {'friends': [{...}], 'id': 9, 'name': 'Klein'}], 'id': 8, 'name': 'Kate'}], 'id': 7, 'name': 'Devin'}], 'id': 5, 'name': 'Clive'}], 'id': 4, 'name': 'Thor'}], 'id': 3, 'name': 'Chi'}], 'id': 2, 'name': 'Sue'}], 'id': 0, 'name': 'Hero'}\n" ] ], [ [ "# 2) what’s the average number of connections", "_____no_output_____" ], [ "Once each user dict contains a list of friends, we can easily ask questions of our\ngraph, like “what’s the average number of connections?”\nFirst we find the total number of connections, by summing up the lengths of all the\nfriends lists:", "_____no_output_____" ] ], [ [ "def number_of_friends(user):\n \"\"\"how many friends does _user_ have?\"\"\"\n return len(user[\"friends\"]) # length of friend_ids list\n\ntotal_connections = sum(number_of_friends(user)\n for user in users) # 24\nprint total_connections", "24\n" ], [ "# And then we just divide by the number of users:\nfrom __future__ import division # integer division is lame\n\nnum_users = len(users) # length of the users list\nprint num_users\navg_connections = total_connections / num_users # 2.4\nprint avg_connections", "10\n2.4\n" ] ], [ [ "It’s also easy to find the most connected people—they’re the people who have the largest\nnumber of friends.\n\nSince there aren’t very many users, we can sort them from “most friends” to “least\nfriends”:", "_____no_output_____" ] ], [ [ "# create a list (user_id, number_of_friends)\nnum_friends_by_id = [(user[\"id\"], number_of_friends(user))\n for user in users]\nprint num_friends_by_id\n\nsorted(num_friends_by_id, # get it sorted\n key=lambda (user_id, num_friends): num_friends, # by num_friends\n reverse=True) # largest to smallest\n\n\n# each pair is (user_id, num_friends)\n# [(1, 3), (2, 3), (3, 3), (5, 3), (8, 3),\n# (0, 2), (4, 2), (6, 2), (7, 2), (9, 1)]", "[(0, 2), (1, 3), (2, 3), (3, 3), (4, 2), (5, 3), (6, 2), (7, 2), (8, 3), (9, 1)]\n" ] ], [ [ "# 3) Data Scientists You May Know", "_____no_output_____" ], [ "Your first instinct is to suggest that a user might know the friends of friends. These\nare easy to compute: for each of a user’s friends, iterate over that person’s friends, and\ncollect all the results:", "_____no_output_____" ] ], [ [ "def friends_of_friend_ids_bad(user):\n # \"foaf\" is short for \"friend of a friend\"\n return [foaf[\"id\"]\n for friend in user[\"friends\"] # for each of user's friends\n for foaf in friend[\"friends\"]] # get each of _their_ friends", "_____no_output_____" ], [ "friends_of_friend_ids_bad(users[0])", "_____no_output_____" ] ], [ [ "It includes user 0 (twice), since Hero is indeed friends with both of his friends. It\nincludes users 1 and 2, although they are both friends with Hero already. And it\nincludes user 3 twice, as Chi is reachable through two different friends:", "_____no_output_____" ] ], [ [ "print [friend[\"id\"] for friend in users[0][\"friends\"]] # [1, 2]\nprint [friend[\"id\"] for friend in users[1][\"friends\"]] # [0, 2, 3]\nprint [friend[\"id\"] for friend in users[2][\"friends\"]] # [0, 1, 3]", "[1, 2]\n[0, 2, 3]\n[0, 1, 3]\n" ] ], [ [ "Knowing that people are friends-of-friends in multiple ways seems like interesting\ninformation, so maybe instead we should produce a count of mutual friends. And we\ndefinitely should use a helper function to exclude people already known to the user:", "_____no_output_____" ] ], [ [ "from collections import Counter # not loaded by default\n\ndef not_the_same(user, other_user):\n \"\"\"two users are not the same if they have different ids\"\"\"\n return user[\"id\"] != other_user[\"id\"]\n\ndef not_friends(user, other_user):\n \"\"\"other_user is not a friend if he's not in user[\"friends\"];\n that is, if he's not_the_same as all the people in user[\"friends\"]\"\"\"\n return all(not_the_same(friend, other_user)\n for friend in user[\"friends\"])\n\ndef friends_of_friend_ids(user):\n return Counter(foaf[\"id\"]\n for friend in user[\"friends\"] # for each of my friends\n for foaf in friend[\"friends\"] # count *their* friends\n if not_the_same(user, foaf) # who aren't me\n and not_friends(user, foaf)) # and aren't my friends\n\nprint friends_of_friend_ids(users[3]) # Counter({0: 2, 5: 1})", "Counter({0: 2, 5: 1})\n" ] ], [ [ "This correctly tells Chi (id 3) that she has two mutual friends with Hero (id 0) but\nonly one mutual friend with Clive (id 5).", "_____no_output_____" ], [ "As a data scientist, you know that you also might enjoy meeting users with similar\ninterests. (This is a good example of the “substantive expertise” aspect of data science.)\nAfter asking around, you manage to get your hands on this data, as a list of\npairs (user_id, interest):", "_____no_output_____" ] ], [ [ "interests = [\n(0, \"Hadoop\"), (0, \"Big Data\"), (0, \"HBase\"), (0, \"Java\"),\n(0, \"Spark\"), (0, \"Storm\"), (0, \"Cassandra\"),\n(1, \"NoSQL\"), (1, \"MongoDB\"), (1, \"Cassandra\"), (1, \"HBase\"),\n(1, \"Postgres\"), (2, \"Python\"), (2, \"scikit-learn\"), (2, \"scipy\"),\n(2, \"numpy\"), (2, \"statsmodels\"), (2, \"pandas\"), (3, \"R\"), (3, \"Python\"),\n(3, \"statistics\"), (3, \"regression\"), (3, \"probability\"),\n(4, \"machine learning\"), (4, \"regression\"), (4, \"decision trees\"),\n(4, \"libsvm\"), (5, \"Python\"), (5, \"R\"), (5, \"Java\"), (5, \"C++\"),\n(5, \"Haskell\"), (5, \"programming languages\"), (6, \"statistics\"),\n(6, \"probability\"), (6, \"mathematics\"), (6, \"theory\"),\n(7, \"machine learning\"), (7, \"scikit-learn\"), (7, \"Mahout\"),\n(7, \"neural networks\"), (8, \"neural networks\"), (8, \"deep learning\"),\n(8, \"Big Data\"), (8, \"artificial intelligence\"), (9, \"Hadoop\"),\n(9, \"Java\"), (9, \"MapReduce\"), (9, \"Big Data\")\n]", "_____no_output_____" ] ], [ [ "For example, Thor (id 4) has no friends in common with Devin (id 7), but they share\nan interest in machine learning.\n\nIt’s easy to build a function that finds users with a certain interest:", "_____no_output_____" ] ], [ [ "def data_scientists_who_like(target_interest):\n return [user_id\n for user_id, user_interest in interests\n if user_interest == target_interest]\n\ndata_scientists_who_like('Java')", "_____no_output_____" ] ], [ [ "This works, but it has to examine the whole list of interests for every search. If we\nhave a lot of users and interests (or if we just want to do a lot of searches), we’re probably\nbetter off building an index from interests to users:", "_____no_output_____" ] ], [ [ "from collections import defaultdict\n\n# keys are interests, values are lists of user_ids with that interest\nuser_ids_by_interest = defaultdict(list)\n\nfor user_id, interest in interests:\n user_ids_by_interest[interest].append(user_id)\n\nprint user_ids_by_interest", "defaultdict(<type 'list'>, {'Java': [0, 5, 9], 'neural networks': [7, 8], 'NoSQL': [1], 'Hadoop': [0, 9], 'Mahout': [7], 'Storm': [0], 'regression': [3, 4], 'statistics': [3, 6], 'probability': [3, 6], 'programming languages': [5], 'Python': [2, 3, 5], 'deep learning': [8], 'Haskell': [5], 'mathematics': [6], 'Spark': [0], 'numpy': [2], 'pandas': [2], 'artificial intelligence': [8], 'theory': [6], 'libsvm': [4], 'C++': [5], 'R': [3, 5], 'HBase': [0, 1], 'Postgres': [1], 'decision trees': [4], 'Big Data': [0, 8, 9], 'MongoDB': [1], 'scikit-learn': [2, 7], 'MapReduce': [9], 'machine learning': [4, 7], 'scipy': [2], 'statsmodels': [2], 'Cassandra': [0, 1]})\n" ], [ "# And another from users to interests:\n# keys are user_ids, values are lists of interests for that user_id\ninterests_by_user_id = defaultdict(list)\n\nfor user_id, interest in interests:\n interests_by_user_id[user_id].append(interest)\n \nprint interests_by_user_id", "defaultdict(<type 'list'>, {0: ['Hadoop', 'Big Data', 'HBase', 'Java', 'Spark', 'Storm', 'Cassandra'], 1: ['NoSQL', 'MongoDB', 'Cassandra', 'HBase', 'Postgres'], 2: ['Python', 'scikit-learn', 'scipy', 'numpy', 'statsmodels', 'pandas'], 3: ['R', 'Python', 'statistics', 'regression', 'probability'], 4: ['machine learning', 'regression', 'decision trees', 'libsvm'], 5: ['Python', 'R', 'Java', 'C++', 'Haskell', 'programming languages'], 6: ['statistics', 'probability', 'mathematics', 'theory'], 7: ['machine learning', 'scikit-learn', 'Mahout', 'neural networks'], 8: ['neural networks', 'deep learning', 'Big Data', 'artificial intelligence'], 9: ['Hadoop', 'Java', 'MapReduce', 'Big Data']})\n" ] ], [ [ "Now it’s easy to find who has the most interests in common with a given user:\n- Iterate over the user’s interests.\n- For each interest, iterate over the other users with that interest.\n- Keep count of how many times we see each other user.", "_____no_output_____" ] ], [ [ "def most_common_interests_with(user):\n return Counter(interested_user_id\n for interest in interests_by_user_id[user[\"id\"]]\n for interested_user_id in user_ids_by_interest[interest]\n if interested_user_id != user[\"id\"])", "_____no_output_____" ] ], [ [ "# 4) Salaries and Experience", "_____no_output_____" ] ], [ [ "# Salary data is of course sensitive,\n# but he manages to provide you an anonymous data set containing each user’s\n# salary (in dollars) and tenure as a data scientist (in years):\n \nsalaries_and_tenures = [(83000, 8.7), (88000, 8.1),\n (48000, 0.7), (76000, 6),\n (69000, 6.5), (76000, 7.5),\n (60000, 2.5), (83000, 10),\n (48000, 1.9), (63000, 4.2)]", "_____no_output_____" ] ], [ [ "It seems pretty clear that people with more experience tend to earn more. How can\nyou turn this into a fun fact? Your first idea is to look at the average salary for each\ntenure:", "_____no_output_____" ] ], [ [ "# keys are years, values are lists of the salaries for each tenure\nsalary_by_tenure = defaultdict(list)\n\nfor salary, tenure in salaries_and_tenures:\n salary_by_tenure[tenure].append(salary)\n \nprint salary_by_tenure\n\n# keys are years, each value is average salary for that tenure\naverage_salary_by_tenure = {\n tenure : sum(salaries) / len(salaries)\n for tenure, salaries in salary_by_tenure.items()\n}\n\nprint average_salary_by_tenure", "defaultdict(<type 'list'>, {6.5: [69000], 7.5: [76000], 6: [76000], 10: [83000], 8.1: [88000], 4.2: [63000], 0.7: [48000], 8.7: [83000], 1.9: [48000], 2.5: [60000]})\n{6.5: 69000.0, 7.5: 76000.0, 6: 76000.0, 10: 83000.0, 8.1: 88000.0, 4.2: 63000.0, 8.7: 83000.0, 0.7: 48000.0, 1.9: 48000.0, 2.5: 60000.0}\n" ] ], [ [ "This turns out to be not particularly useful, as none of the users have the same tenure, which means we’re just reporting the individual users’ salaries.", "_____no_output_____" ] ], [ [ "# It might be more helpful to bucket the tenures:\ndef tenure_bucket(tenure):\n if tenure < 2:\n return \"less than two\"\n elif tenure < 5:\n return \"between two and five\"\n else:\n return \"more than five\"", "_____no_output_____" ], [ "# Then group together the salaries corresponding to each bucket:\n\n# keys are tenure buckets, values are lists of salaries for that bucket\nsalary_by_tenure_bucket = defaultdict(list)\n\nfor salary, tenure in salaries_and_tenures:\n bucket = tenure_bucket(tenure)\n salary_by_tenure_bucket[bucket].append(salary)\n \n# And finally compute the average salary for each group:\n\n# keys are tenure buckets, values are average salary for that bucket\naverage_salary_by_bucket = {\n tenure_bucket : sum(salaries) / len(salaries)\n for tenure_bucket, salaries in salary_by_tenure_bucket.iteritems()\n}\n\nprint average_salary_by_bucket", "{'more than five': 79166.66666666667, 'between two and five': 61500.0, 'less than two': 48000.0}\n" ] ], [ [ "# 5) Paid Accounts", "_____no_output_____" ] ], [ [ "def predict_paid_or_unpaid(years_experience):\n if years_experience < 3.0:\n return \"paid\"\n elif years_experience < 8.5:\n return \"unpaid\"\n else:\n return \"paid\"", "_____no_output_____" ] ], [ [ "# 6) Topics of Interest", "_____no_output_____" ], [ "One simple (if not particularly exciting) way to find the most popular interests is simply\nto count the words:\n1. Lowercase each interest (since different users may or may not capitalize their\ninterests).\n2. Split it into words.\n3. Count the results.", "_____no_output_____" ] ], [ [ "words_and_counts = Counter(word\n for user, interest in interests\n for word in interest.lower().split())\n\nprint words_and_counts", "Counter({'learning': 3, 'java': 3, 'python': 3, 'big': 3, 'data': 3, 'hbase': 2, 'regression': 2, 'cassandra': 2, 'statistics': 2, 'probability': 2, 'hadoop': 2, 'networks': 2, 'machine': 2, 'neural': 2, 'scikit-learn': 2, 'r': 2, 'nosql': 1, 'programming': 1, 'deep': 1, 'haskell': 1, 'languages': 1, 'decision': 1, 'artificial': 1, 'storm': 1, 'mongodb': 1, 'intelligence': 1, 'mathematics': 1, 'numpy': 1, 'pandas': 1, 'postgres': 1, 'libsvm': 1, 'trees': 1, 'scipy': 1, 'spark': 1, 'mapreduce': 1, 'c++': 1, 'theory': 1, 'statsmodels': 1, 'mahout': 1})\n" ] ], [ [ "This makes it easy to list out the words that occur more than once:", "_____no_output_____" ] ], [ [ "for word, count in words_and_counts.most_common():\n if count > 1:\n print word, count", "learning 3\njava 3\npython 3\nbig 3\ndata 3\nhbase 2\nregression 2\ncassandra 2\nstatistics 2\nprobability 2\nhadoop 2\nnetworks 2\nmachine 2\nneural 2\nscikit-learn 2\nr 2\n" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import math\n\nfrom IPython import display\nfrom matplotlib import cm\nfrom matplotlib import gridspec\nfrom matplotlib import pyplot as plt\nimport numpy as np\nimport pandas as pd\nfrom sklearn import metrics\nimport tensorflow as tf\nfrom tensorflow.python.data import Dataset\n\ntf.logging.set_verbosity(tf.logging.ERROR)\npd.options.display.max_rows = 10\npd.options.display.float_format = '{:.1f}'.format\n\nlottery_dataframe = pd.read_csv(\"lottery.csv\", sep=\",\")\nlottery_dataframe = lottery_dataframe.reindex(np.random.permutation(lottery_dataframe.index))", "_____no_output_____" ], [ "def preprocess_features(lottery_dataframe):\n selected_features = lottery_dataframe[\n [\"day\",\n \"month\",\n \"year\",\n #\"no1\",\n #\"top2\",\n #\"top3\",\n #\"front3_1\",\n #\"front3_2\",\n #\"bottom3_1\", \n #\"bottom3_2\"\n ]]\n return selected_features\n\ndef preprocess_targets(lottery_dataframe):\n output_targets = pd.DataFrame()\n output_targets[\"bottom2\"] = lottery_dataframe[\"bottom2\"]\n return output_targets", "_____no_output_____" ], [ "# Choose the first 578 (out of 684) examples for training.\ntraining_examples = preprocess_features(lottery_dataframe.head(578))\ntraining_targets = preprocess_targets(lottery_dataframe.head(578))\n\n# Choose the last 106 (out of 684) examples for validation.\nvalidation_examples = preprocess_features(lottery_dataframe.tail(106))\nvalidation_targets = preprocess_targets(lottery_dataframe.tail(106))\n\n# Double-check that we've done the right thing.\nprint(\"Training examples summary:\")\ndisplay.display(training_examples.describe())\nprint(\"Validation examples summary:\")\ndisplay.display(validation_examples.describe())\n\nprint(\"Training targets summary:\")\ndisplay.display(training_targets.describe())\nprint(\"Validation targets summary:\")\ndisplay.display(validation_targets.describe())", "Training examples summary:\n" ], [ "def construct_feature_columns(input_features):\n return set([tf.feature_column.numeric_column(my_feature)\n for my_feature in input_features])", "_____no_output_____" ], [ "def my_input_fn(features, targets, batch_size=1, shuffle=True, num_epochs=None):\n \"\"\"Trains a neural network model.\n \n Args:\n features: pandas DataFrame of features\n targets: pandas DataFrame of targets\n batch_size: Size of batches to be passed to the model\n shuffle: True or False. Whether to shuffle the data.\n num_epochs: Number of epochs for which data should be repeated. None = repeat indefinitely\n Returns:\n Tuple of (features, labels) for next data batch\n \"\"\"\n \n # Convert pandas data into a dict of np arrays.\n features = {key:np.array(value) for key,value in dict(features).items()} \n \n # Construct a dataset, and configure batching/repeating.\n ds = Dataset.from_tensor_slices((features,targets)) # warning: 2GB limit\n ds = ds.batch(batch_size).repeat(num_epochs)\n \n # Shuffle the data, if specified.\n if shuffle:\n ds = ds.shuffle(10000)\n \n # Return the next batch of data.\n features, labels = ds.make_one_shot_iterator().get_next()\n return features, labels", "_____no_output_____" ], [ "def train_model(\n learning_rate,\n steps,\n batch_size,\n feature_columns,\n training_examples,\n training_targets,\n validation_examples,\n validation_targets):\n \"\"\"Trains a linear regression model.\n \n In addition to training, this function also prints training progress information,\n as well as a plot of the training and validation loss over time.\n \n Args:\n learning_rate: A `float`, the learning rate.\n steps: A non-zero `int`, the total number of training steps. A training step\n consists of a forward and backward pass using a single batch.\n feature_columns: A `set` specifying the input feature columns to use.\n training_examples: A `DataFrame` containing one or more columns from\n `california_housing_dataframe` to use as input features for training.\n training_targets: A `DataFrame` containing exactly one column from\n `california_housing_dataframe` to use as target for training.\n validation_examples: A `DataFrame` containing one or more columns from\n `california_housing_dataframe` to use as input features for validation.\n validation_targets: A `DataFrame` containing exactly one column from\n `california_housing_dataframe` to use as target for validation.\n \n Returns:\n A `LinearRegressor` object trained on the training data.\n \"\"\"\n\n periods = 10\n steps_per_period = steps / periods\n\n # Create a linear regressor object.\n my_optimizer = tf.train.FtrlOptimizer(learning_rate=learning_rate)\n my_optimizer = tf.contrib.estimator.clip_gradients_by_norm(my_optimizer, 5.0)\n linear_regressor = tf.estimator.LinearRegressor(\n feature_columns=feature_columns,\n optimizer=my_optimizer\n )\n \n training_input_fn = lambda: my_input_fn(training_examples, \n training_targets[\"bottom2\"], \n batch_size=batch_size)\n predict_training_input_fn = lambda: my_input_fn(training_examples, \n training_targets[\"bottom2\"], \n num_epochs=1, \n shuffle=False)\n predict_validation_input_fn = lambda: my_input_fn(validation_examples, \n validation_targets[\"bottom2\"], \n num_epochs=1, \n shuffle=False)\n\n # Train the model, but do so inside a loop so that we can periodically assess\n # loss metrics.\n print(\"Training model...\")\n print(\"RMSE (on training data):\")\n training_rmse = []\n validation_rmse = []\n for period in range (0, periods):\n # Train the model, starting from the prior state.\n linear_regressor.train(\n input_fn=training_input_fn,\n steps=steps_per_period\n )\n # Take a break and compute predictions.\n training_predictions = linear_regressor.predict(input_fn=predict_training_input_fn)\n training_predictions = np.array([item['predictions'][0] for item in training_predictions])\n validation_predictions = linear_regressor.predict(input_fn=predict_validation_input_fn)\n validation_predictions = np.array([item['predictions'][0] for item in validation_predictions])\n \n # Compute training and validation loss.\n training_root_mean_squared_error = math.sqrt(\n metrics.mean_squared_error(training_predictions, training_targets))\n validation_root_mean_squared_error = math.sqrt(\n metrics.mean_squared_error(validation_predictions, validation_targets))\n # Occasionally print the current loss.\n print(\" period %02d : %0.2f\" % (period, training_root_mean_squared_error))\n # Add the loss metrics from this period to our list.\n training_rmse.append(training_root_mean_squared_error)\n validation_rmse.append(validation_root_mean_squared_error)\n print(\"Model training finished.\")\n\n \n # Output a graph of loss metrics over periods.\n plt.ylabel(\"RMSE\")\n plt.xlabel(\"Periods\")\n plt.title(\"Root Mean Squared Error vs. Periods\")\n plt.tight_layout()\n plt.plot(training_rmse, label=\"training\")\n plt.plot(validation_rmse, label=\"validation\")\n plt.legend()\n\n return linear_regressor, validation_predictions", "_____no_output_____" ], [ "r, p = train_model(\n learning_rate=0.001,\n steps=5000,\n batch_size=10,\n feature_columns=construct_feature_columns(training_examples),\n training_examples=training_examples,\n training_targets=training_targets,\n validation_examples=validation_examples,\n validation_targets=validation_targets)", "Training model...\nRMSE (on training data):\n period 00 : 28.87\n period 01 : 28.87\n period 02 : 28.87\n period 03 : 28.87\n period 04 : 28.87\n period 05 : 28.88\n period 06 : 28.87\n period 07 : 28.87\n period 08 : 28.87\n period 09 : 28.87\nModel training finished.\n" ], [ "my_target = validation_targets.copy()\nmy_target['predict'] = p\nmy_target\n\n# Output a graph of loss metrics over periods.\nplt.ylabel(\"target\")\nplt.xlabel(\"predict\")\nplt.title(\"bottom2\")\nplt.tight_layout()\nplt.plot(validation_targets, label=\"Target\")\nplt.plot(p, label=\"Predict\")\nplt.legend()\n\nmy_target", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "import pandas as pd\nimport numpy as np\nimport matplotlib.pyplot as plt", "_____no_output_____" ], [ "import pandas as pd", "_____no_output_____" ], [ "data=pd.read_csv(\"Social_Network_Ads.csv\")", "_____no_output_____" ], [ "data", "_____no_output_____" ], [ "data.head()", "_____no_output_____" ], [ "data.tail()", "_____no_output_____" ], [ "data.dropna(axis=0,inplace=True)", "_____no_output_____" ], [ "data.shape", "_____no_output_____" ], [ "data.head(5)", "_____no_output_____" ], [ "plt.scatter(data['Age'],data['EstimatedSalary'],c='blue')\nplt.xlabel('Age(months)')\nplt.ylabel('EstimatedSalary')\nplt.show()", "_____no_output_____" ], [ "plt.hist(data['Gender'])\nplt.show()", "_____no_output_____" ], [ "plt.hist(data['EstimatedSalary'])", "_____no_output_____" ], [ "plt.hist(data['Age'],color='orange',edgecolor='white',bins=5)\nplt.title=(\"histogram of ages\")\nplt.xlabel(\"frequencies\")\nplt.ylabel(\"Age\")\nplt.show()", "_____no_output_____" ], [ "import seaborn as sns", "_____no_output_____" ], [ "sns.set(style='darkgrid')\nsns.regplot(x=data['Age'],y=data['EstimatedSalary'])", "_____no_output_____" ], [ "sns.set(style='darkgrid')\nsns.regplot(x=data['Age'],y=data['EstimatedSalary'],marker='*',fit_reg=False)", "_____no_output_____" ], [ "sns.distplot(data['Age'])", "_____no_output_____" ], [ "sns.distplot(data['Age'],kde=False)", "_____no_output_____" ], [ "sns.distplot(data['Age'],kde=False,bins=5)", "_____no_output_____" ], [ "sns.countplot(x=\"Purchased\",data=data)", "_____no_output_____" ], [ "sns.boxplot(y=data['EstimatedSalary'])", "_____no_output_____" ], [ "sns.b", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# LOFO Feature Importance\nhttps://github.com/aerdem4/lofo-importance", "_____no_output_____" ] ], [ [ "!pip install lofo-importance", "Collecting lofo-importance\r\n Downloading https://files.pythonhosted.org/packages/43/8e/c43fb5e6a56e6e028b24c93bee6282726534abeb42c1cd12d41fecd5443b/lofo_importance-0.2.0-py3-none-any.whl\r\nRequirement already satisfied: lightgbm in /opt/conda/lib/python3.6/site-packages (from lofo-importance) (2.2.3)\r\nRequirement already satisfied: matplotlib in /opt/conda/lib/python3.6/site-packages (from lofo-importance) (3.0.3)\r\nRequirement already satisfied: scikit-learn in /opt/conda/lib/python3.6/site-packages (from lofo-importance) (0.20.3)\r\nRequirement already satisfied: numpy in /opt/conda/lib/python3.6/site-packages (from lofo-importance) (1.16.3)\r\nRequirement already satisfied: ipywidgets in /opt/conda/lib/python3.6/site-packages (from lofo-importance) (7.2.1)\r\nRequirement already satisfied: jupyter in /opt/conda/lib/python3.6/site-packages (from lofo-importance) (1.0.0)\r\nRequirement already satisfied: tqdm in /opt/conda/lib/python3.6/site-packages (from lofo-importance) (4.31.1)\r\nRequirement already satisfied: pandas in /opt/conda/lib/python3.6/site-packages (from lofo-importance) (0.23.4)\r\nRequirement already satisfied: scipy in /opt/conda/lib/python3.6/site-packages (from lightgbm->lofo-importance) (1.1.0)\r\nRequirement already satisfied: python-dateutil>=2.1 in /opt/conda/lib/python3.6/site-packages (from matplotlib->lofo-importance) (2.6.0)\r\nRequirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /opt/conda/lib/python3.6/site-packages (from matplotlib->lofo-importance) (2.2.0)\r\nRequirement already satisfied: cycler>=0.10 in /opt/conda/lib/python3.6/site-packages (from matplotlib->lofo-importance) (0.10.0)\r\nRequirement already satisfied: kiwisolver>=1.0.1 in /opt/conda/lib/python3.6/site-packages (from matplotlib->lofo-importance) (1.0.1)\r\nRequirement already satisfied: ipykernel>=4.5.1 in /opt/conda/lib/python3.6/site-packages (from ipywidgets->lofo-importance) (4.8.2)\r\nRequirement already satisfied: traitlets>=4.3.1 in /opt/conda/lib/python3.6/site-packages (from ipywidgets->lofo-importance) (4.3.2)\r\nRequirement already satisfied: nbformat>=4.2.0 in /opt/conda/lib/python3.6/site-packages (from ipywidgets->lofo-importance) (4.4.0)\r\nRequirement already satisfied: widgetsnbextension~=3.2.0 in /opt/conda/lib/python3.6/site-packages (from ipywidgets->lofo-importance) (3.2.1)\r\nRequirement already satisfied: ipython>=4.0.0 in /opt/conda/lib/python3.6/site-packages (from ipywidgets->lofo-importance) (6.4.0)\r\nRequirement already satisfied: notebook in /opt/conda/lib/python3.6/site-packages (from jupyter->lofo-importance) (5.5.0)\r\nRequirement already satisfied: qtconsole in /opt/conda/lib/python3.6/site-packages (from jupyter->lofo-importance) (4.3.1)\r\nRequirement already satisfied: jupyter-console in /opt/conda/lib/python3.6/site-packages (from jupyter->lofo-importance) (5.2.0)\r\nRequirement already satisfied: nbconvert in /opt/conda/lib/python3.6/site-packages (from jupyter->lofo-importance) (5.3.1)\r\nRequirement already satisfied: pytz>=2011k in /opt/conda/lib/python3.6/site-packages (from pandas->lofo-importance) (2018.4)\r\nRequirement already satisfied: six>=1.5 in /opt/conda/lib/python3.6/site-packages (from python-dateutil>=2.1->matplotlib->lofo-importance) (1.12.0)\r\nRequirement already satisfied: setuptools in /opt/conda/lib/python3.6/site-packages (from kiwisolver>=1.0.1->matplotlib->lofo-importance) (39.1.0)\r\nRequirement already satisfied: jupyter_client in /opt/conda/lib/python3.6/site-packages (from ipykernel>=4.5.1->ipywidgets->lofo-importance) (5.2.3)\r\nRequirement already satisfied: tornado>=4.0 in /opt/conda/lib/python3.6/site-packages (from ipykernel>=4.5.1->ipywidgets->lofo-importance) (5.0.2)\r\nRequirement already satisfied: ipython_genutils in /opt/conda/lib/python3.6/site-packages (from traitlets>=4.3.1->ipywidgets->lofo-importance) (0.2.0)\r\nRequirement already satisfied: decorator in /opt/conda/lib/python3.6/site-packages (from traitlets>=4.3.1->ipywidgets->lofo-importance) (4.3.0)\r\nRequirement already satisfied: jsonschema!=2.5.0,>=2.4 in /opt/conda/lib/python3.6/site-packages (from nbformat>=4.2.0->ipywidgets->lofo-importance) (2.6.0)\r\nRequirement already satisfied: jupyter_core in /opt/conda/lib/python3.6/site-packages (from nbformat>=4.2.0->ipywidgets->lofo-importance) (4.4.0)\r\nRequirement already satisfied: prompt-toolkit<2.0.0,>=1.0.15 in /opt/conda/lib/python3.6/site-packages (from ipython>=4.0.0->ipywidgets->lofo-importance) (1.0.15)\r\nRequirement already satisfied: simplegeneric>0.8 in /opt/conda/lib/python3.6/site-packages (from ipython>=4.0.0->ipywidgets->lofo-importance) (0.8.1)\r\nRequirement already satisfied: pygments in /opt/conda/lib/python3.6/site-packages (from ipython>=4.0.0->ipywidgets->lofo-importance) (2.2.0)\r\nRequirement already satisfied: backcall in /opt/conda/lib/python3.6/site-packages (from ipython>=4.0.0->ipywidgets->lofo-importance) (0.1.0)\r\nRequirement already satisfied: pexpect; sys_platform != \"win32\" in /opt/conda/lib/python3.6/site-packages (from ipython>=4.0.0->ipywidgets->lofo-importance) (4.5.0)\r\nRequirement already satisfied: jedi>=0.10 in /opt/conda/lib/python3.6/site-packages (from ipython>=4.0.0->ipywidgets->lofo-importance) (0.12.0)\r\nRequirement already satisfied: pickleshare in /opt/conda/lib/python3.6/site-packages (from ipython>=4.0.0->ipywidgets->lofo-importance) (0.7.4)\r\nRequirement already satisfied: Send2Trash in /opt/conda/lib/python3.6/site-packages (from notebook->jupyter->lofo-importance) (1.5.0)\r\nRequirement already satisfied: terminado>=0.8.1 in /opt/conda/lib/python3.6/site-packages (from notebook->jupyter->lofo-importance) (0.8.1)\r\nRequirement already satisfied: jinja2 in /opt/conda/lib/python3.6/site-packages (from notebook->jupyter->lofo-importance) (2.10)\r\nRequirement already satisfied: pyzmq>=17 in /opt/conda/lib/python3.6/site-packages (from notebook->jupyter->lofo-importance) (17.0.0)\r\nRequirement already satisfied: mistune>=0.7.4 in /opt/conda/lib/python3.6/site-packages (from nbconvert->jupyter->lofo-importance) (0.8.3)\r\nRequirement already satisfied: entrypoints>=0.2.2 in /opt/conda/lib/python3.6/site-packages (from nbconvert->jupyter->lofo-importance) (0.2.3)\r\nRequirement already satisfied: bleach in /opt/conda/lib/python3.6/site-packages (from nbconvert->jupyter->lofo-importance) (2.1.3)\r\nRequirement already satisfied: pandocfilters>=1.4.1 in /opt/conda/lib/python3.6/site-packages (from nbconvert->jupyter->lofo-importance) (1.4.2)\r\nRequirement already satisfied: testpath in /opt/conda/lib/python3.6/site-packages (from nbconvert->jupyter->lofo-importance) (0.3.1)\r\nRequirement already satisfied: wcwidth in /opt/conda/lib/python3.6/site-packages (from prompt-toolkit<2.0.0,>=1.0.15->ipython>=4.0.0->ipywidgets->lofo-importance) (0.1.7)\r\nRequirement already satisfied: ptyprocess>=0.5 in /opt/conda/lib/python3.6/site-packages (from pexpect; sys_platform != \"win32\"->ipython>=4.0.0->ipywidgets->lofo-importance) (0.5.2)\r\nRequirement already satisfied: parso>=0.2.0 in /opt/conda/lib/python3.6/site-packages (from jedi>=0.10->ipython>=4.0.0->ipywidgets->lofo-importance) (0.2.0)\r\nRequirement already satisfied: MarkupSafe>=0.23 in /opt/conda/lib/python3.6/site-packages (from jinja2->notebook->jupyter->lofo-importance) (1.0)\r\nRequirement already satisfied: html5lib!=1.0b1,!=1.0b2,!=1.0b3,!=1.0b4,!=1.0b5,!=1.0b6,!=1.0b7,!=1.0b8,>=0.99999999pre in /opt/conda/lib/python3.6/site-packages (from bleach->nbconvert->jupyter->lofo-importance) (1.0.1)\r\nRequirement already satisfied: webencodings in /opt/conda/lib/python3.6/site-packages (from html5lib!=1.0b1,!=1.0b2,!=1.0b3,!=1.0b4,!=1.0b5,!=1.0b6,!=1.0b7,!=1.0b8,>=0.99999999pre->bleach->nbconvert->jupyter->lofo-importance) (0.5.1)\r\nInstalling collected packages: lofo-importance\r\nSuccessfully installed lofo-importance-0.2.0\r\n\u001b[33mYou are using pip version 19.0.3, however version 19.1.1 is available.\r\nYou should consider upgrading via the 'pip install --upgrade pip' command.\u001b[0m\r\n" ], [ "import numpy as np\nimport pandas as pd\n\ndf = pd.read_csv(\"../input/train.csv\", index_col='id')\ndf['wheezy-copper-turtle-magic'] = df['wheezy-copper-turtle-magic'].astype('category')\ndf.shape", "_____no_output_____" ] ], [ [ "### Use the best model in public kernels", "_____no_output_____" ] ], [ [ "from sklearn.pipeline import Pipeline\nfrom sklearn.preprocessing import StandardScaler\nfrom sklearn.discriminant_analysis import QuadraticDiscriminantAnalysis\n\ndef get_model():\n return Pipeline([('scaler', StandardScaler()),\n ('qda', QuadraticDiscriminantAnalysis(reg_param=0.111))\n ])", "_____no_output_____" ] ], [ [ "### Top 20 Features for wheezy-copper-turtle-magic = 0", "_____no_output_____" ] ], [ [ "from sklearn.model_selection import KFold, StratifiedKFold, train_test_split\nfrom sklearn.linear_model import LogisticRegression\nfrom lofo import LOFOImportance, FLOFOImportance, plot_importance\n\n\nfeatures = [c for c in df.columns if c not in ['id', 'target', 'wheezy-copper-turtle-magic']]\n\n\ndef get_lofo_importance(wctm_num):\n sub_df = df[df['wheezy-copper-turtle-magic'] == wctm_num]\n sub_features = [f for f in features if sub_df[f].std() > 1.5]\n lofo_imp = LOFOImportance(sub_df, target=\"target\",\n features=sub_features, \n cv=StratifiedKFold(n_splits=4, random_state=42, shuffle=True), scoring=\"roc_auc\",\n model=get_model(), n_jobs=4)\n return lofo_imp.get_importance()\n\nplot_importance(get_lofo_importance(0), figsize=(12, 12))", "/opt/conda/lib/python3.6/site-packages/lofo/lofo_importance.py:32: UserWarning: Warning: If your model is multithreaded, please initialise the number of jobs of LOFO to be equal to 1, otherwise you may experience issues.\n warnings.warn(warning_str)\n" ] ], [ [ "### Top 20 Features for wheezy-copper-turtle-magic = 1", "_____no_output_____" ] ], [ [ "plot_importance(get_lofo_importance(1), figsize=(12, 12))", "/opt/conda/lib/python3.6/site-packages/lofo/lofo_importance.py:32: UserWarning: Warning: If your model is multithreaded, please initialise the number of jobs of LOFO to be equal to 1, otherwise you may experience issues.\n warnings.warn(warning_str)\n" ] ], [ [ "### Top 20 Features for wheezy-copper-turtle-magic = 2", "_____no_output_____" ] ], [ [ "plot_importance(get_lofo_importance(2), figsize=(12, 12))", "/opt/conda/lib/python3.6/site-packages/lofo/lofo_importance.py:32: UserWarning: Warning: If your model is multithreaded, please initialise the number of jobs of LOFO to be equal to 1, otherwise you may experience issues.\n warnings.warn(warning_str)\n" ] ], [ [ "### Find the most harmful features for each wheezy-copper-turtle-magic", "_____no_output_____" ] ], [ [ "from tqdm import tqdm_notebook\nimport warnings\nwarnings.filterwarnings(\"ignore\")\n\nfeatures_to_remove = []\npotential_gain = []\n\nfor i in tqdm_notebook(range(512)):\n imp = get_lofo_importance(i)\n features_to_remove.append(imp[\"feature\"].values[-1])\n potential_gain.append(-imp[\"importance_mean\"].values[-1])\n \nprint(\"Potential gain (AUC):\", np.round(np.mean(potential_gain), 5))", "_____no_output_____" ], [ "features_to_remove", "_____no_output_____" ] ], [ [ "# Create submission using the current best kernel\nhttps://www.kaggle.com/tunguz/ig-pca-nusvc-knn-qda-lr-stack by Bojan Tunguz", "_____no_output_____" ] ], [ [ "import numpy as np, pandas as pd\nfrom sklearn.model_selection import StratifiedKFold\nfrom sklearn.metrics import roc_auc_score\nfrom sklearn import svm, neighbors, linear_model, neural_network\nfrom sklearn.svm import NuSVC\nfrom sklearn.preprocessing import StandardScaler\nfrom sklearn.decomposition import PCA\nfrom tqdm import tqdm\nfrom sklearn.discriminant_analysis import QuadraticDiscriminantAnalysis\nfrom sklearn.pipeline import Pipeline\nfrom sklearn.metrics import roc_auc_score\nfrom sklearn.feature_selection import VarianceThreshold\n\ntrain = pd.read_csv('../input/train.csv')\ntest = pd.read_csv('../input/test.csv')\n\noof_svnu = np.zeros(len(train)) \npred_te_svnu = np.zeros(len(test))\n\noof_svc = np.zeros(len(train)) \npred_te_svc = np.zeros(len(test))\n\noof_knn = np.zeros(len(train)) \npred_te_knn = np.zeros(len(test))\n\noof_lr = np.zeros(len(train)) \npred_te_lr = np.zeros(len(test))\n\noof_mlp = np.zeros(len(train)) \npred_te_mlp = np.zeros(len(test))\n\noof_qda = np.zeros(len(train)) \npred_te_qda = np.zeros(len(test))\n\ndefault_cols = [c for c in train.columns if c not in ['id', 'target', 'wheezy-copper-turtle-magic']]\n\nfor i in range(512):\n cols = [c for c in default_cols if c != features_to_remove[i]]\n \n train2 = train[train['wheezy-copper-turtle-magic']==i]\n test2 = test[test['wheezy-copper-turtle-magic']==i]\n idx1 = train2.index; idx2 = test2.index\n train2.reset_index(drop=True,inplace=True)\n\n data = pd.concat([pd.DataFrame(train2[cols]), pd.DataFrame(test2[cols])])\n data2 = StandardScaler().fit_transform(PCA(svd_solver='full',n_components='mle').fit_transform(data[cols]))\n train3 = data2[:train2.shape[0]]; test3 = data2[train2.shape[0]:]\n \n data2 = StandardScaler().fit_transform(VarianceThreshold(threshold=1.5).fit_transform(data[cols]))\n train4 = data2[:train2.shape[0]]; test4 = data2[train2.shape[0]:]\n \n # STRATIFIED K FOLD (Using splits=25 scores 0.002 better but is slower)\n skf = StratifiedKFold(n_splits=5, random_state=42)\n for train_index, test_index in skf.split(train2, train2['target']):\n\n clf = NuSVC(probability=True, kernel='poly', degree=4, gamma='auto', random_state=4, nu=0.59, coef0=0.053)\n clf.fit(train3[train_index,:],train2.loc[train_index]['target'])\n oof_svnu[idx1[test_index]] = clf.predict_proba(train3[test_index,:])[:,1]\n pred_te_svnu[idx2] += clf.predict_proba(test3)[:,1] / skf.n_splits\n \n clf = neighbors.KNeighborsClassifier(n_neighbors=17, p=2.9)\n clf.fit(train3[train_index,:],train2.loc[train_index]['target'])\n oof_knn[idx1[test_index]] = clf.predict_proba(train3[test_index,:])[:,1]\n pred_te_knn[idx2] += clf.predict_proba(test3)[:,1] / skf.n_splits\n \n clf = linear_model.LogisticRegression(solver='saga',penalty='l1',C=0.1)\n clf.fit(train3[train_index,:],train2.loc[train_index]['target'])\n oof_lr[idx1[test_index]] = clf.predict_proba(train3[test_index,:])[:,1]\n pred_te_lr[idx2] += clf.predict_proba(test3)[:,1] / skf.n_splits\n \n clf = neural_network.MLPClassifier(random_state=3, activation='relu', solver='lbfgs', tol=1e-06, hidden_layer_sizes=(250, ))\n clf.fit(train3[train_index,:],train2.loc[train_index]['target'])\n oof_mlp[idx1[test_index]] = clf.predict_proba(train3[test_index,:])[:,1]\n pred_te_mlp[idx2] += clf.predict_proba(test3)[:,1] / skf.n_splits\n \n clf = svm.SVC(probability=True, kernel='poly', degree=4, gamma='auto', random_state=42)\n clf.fit(train3[train_index,:],train2.loc[train_index]['target'])\n oof_svc[idx1[test_index]] = clf.predict_proba(train3[test_index,:])[:,1]\n pred_te_svc[idx2] += clf.predict_proba(test3)[:,1] / skf.n_splits\n \n clf = QuadraticDiscriminantAnalysis(reg_param=0.111)\n clf.fit(train4[train_index,:],train2.loc[train_index]['target'])\n oof_qda[idx1[test_index]] = clf.predict_proba(train4[test_index,:])[:,1]\n pred_te_qda[idx2] += clf.predict_proba(test4)[:,1] / skf.n_splits\n \n \nprint('lr', roc_auc_score(train['target'], oof_lr))\nprint('knn', roc_auc_score(train['target'], oof_knn))\nprint('svc', roc_auc_score(train['target'], oof_svc))\nprint('svcnu', roc_auc_score(train['target'], oof_svnu))\nprint('mlp', roc_auc_score(train['target'], oof_mlp))\nprint('qda', roc_auc_score(train['target'], oof_qda))\nprint('blend 1', roc_auc_score(train['target'], oof_svnu*0.7 + oof_svc*0.05 + oof_knn*0.2 + oof_mlp*0.05))\nprint('blend 2', roc_auc_score(train['target'], oof_qda*0.5+oof_svnu*0.35 + oof_svc*0.025 + oof_knn*0.1 + oof_mlp*0.025))\n\noof_svnu = oof_svnu.reshape(-1, 1)\npred_te_svnu = pred_te_svnu.reshape(-1, 1)\noof_svc = oof_svc.reshape(-1, 1)\npred_te_svc = pred_te_svc.reshape(-1, 1)\noof_knn = oof_knn.reshape(-1, 1)\npred_te_knn = pred_te_knn.reshape(-1, 1)\noof_mlp = oof_mlp.reshape(-1, 1)\npred_te_mlp = pred_te_mlp.reshape(-1, 1)\noof_lr = oof_lr.reshape(-1, 1)\npred_te_lr = pred_te_lr.reshape(-1, 1)\noof_qda = oof_qda.reshape(-1, 1)\npred_te_qda = pred_te_qda.reshape(-1, 1)\n\ntr = np.concatenate((oof_svnu, oof_svc, oof_knn, oof_mlp, oof_lr, oof_qda), axis=1)\nte = np.concatenate((pred_te_svnu, pred_te_svc, pred_te_knn, pred_te_mlp, pred_te_lr, pred_te_qda), axis=1)\nprint(tr.shape, te.shape)\n\noof_lrr = np.zeros(len(train)) \npred_te_lrr = np.zeros(len(test))\nskf = StratifiedKFold(n_splits=5, random_state=42)\n\nfor train_index, test_index in skf.split(tr, train['target']):\n lrr = linear_model.LogisticRegression()\n lrr.fit(tr[train_index], train['target'][train_index])\n oof_lrr[test_index] = lrr.predict_proba(tr[test_index,:])[:,1]\n pred_te_lrr += lrr.predict_proba(te)[:,1] / skf.n_splits\n \nprint('stack CV score =',round(roc_auc_score(train['target'],oof_lrr),6))\n\nsub = pd.read_csv('../input/sample_submission.csv')\n\nsub['target'] = pred_te_lrr\nsub.to_csv('submission_stack.csv', index=False)", "lr 0.790131655722408\nknn 0.9016209110875143\nsvc 0.9496455309107024\nsvcnu 0.9602070184471454\nmlp 0.9099946396711365\nqda 0.9645745359410043\nblend 1 0.9606747834832012\nblend 2 0.9658290716499904\n(262144, 6) (131073, 6)\nstack CV score = 0.965867\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ] ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ [ [ "Submitting various things for end of grant.", "_____no_output_____" ] ], [ [ "import os\nimport sys\nimport requests\nimport pandas\nimport paramiko\nimport json\nfrom IPython import display", "_____no_output_____" ], [ "from curation_common import *\nfrom htsworkflow.submission.encoded import DCCValidator", "_____no_output_____" ], [ "PANDAS_ODF = os.path.expanduser('~/src/odf_pandas')\nif PANDAS_ODF not in sys.path:\n sys.path.append(PANDAS_ODF)\n from pandasodf import ODFReader", "_____no_output_____" ], [ "import gcat", "_____no_output_____" ], [ "from htsworkflow.submission.encoded import Document\nfrom htsworkflow.submission.aws_submission import run_aws_cp", "_____no_output_____" ], [ "# live server & control file\n#server = ENCODED('www.encodeproject.org')\nspreadsheet_name = \"ENCODE_test_miRNA_experiments_01112018\"\n\n# test server & datafile\nserver = ENCODED('test.encodedcc.org')\n#spreadsheet_name = os.path.expanduser('~diane/woldlab/ENCODE/C1-encode3-limb-2017-testserver.ods')\n\nserver.load_netrc()\nvalidator = DCCValidator(server)", "_____no_output_____" ], [ "award = 'UM1HG009443'", "_____no_output_____" ] ], [ [ "# Submit Documents", "_____no_output_____" ], [ "Example Document submission", "_____no_output_____" ] ], [ [ "#atac_uuid = '0fc44318-b802-474e-8199-f3b6d708eb6f'\n#atac = Document(os.path.expanduser('~/proj/encode3-curation/Wold_Lab_ATAC_Seq_protocol_December_2016.pdf'),\n# 'general protocol',\n# 'ATAC-Seq experiment protocol for Wold lab',\n# )\n#body = atac.create_if_needed(server, atac_uuid)\n#print(body['@id'])", "_____no_output_____" ] ], [ [ "# Submit Annotations", "_____no_output_____" ] ], [ [ "#sheet = gcat.get_file(spreadsheet_name, fmt='pandas_excel')\n#annotations = sheet.parse('Annotations', header=0)\n#created = server.post_sheet('/annotations/', annotations, verbose=True, dry_run=True)\n#print(len(created))", "_____no_output_____" ], [ "#if created:\n# annotations.to_excel('/tmp/annotations.xlsx', index=False)", "_____no_output_____" ] ], [ [ "# Register Biosamples", "_____no_output_____" ] ], [ [ "book = gcat.get_file(spreadsheet_name, fmt='pandas_excel')\nbiosample = book.parse('Biosamples', header=0)\ncreated = server.post_sheet('/biosamples/', biosample, \n verbose=True, \n dry_run=True, \n validator=validator)\nprint(len(created))", "_____no_output_____" ], [ "if created:\n biosample.to_excel('/dev/shm/biosamples.xlsx', index=False)", "_____no_output_____" ] ], [ [ "# Register Libraries", "_____no_output_____" ] ], [ [ "print(spreadsheet_name)\nbook = gcat.get_file(spreadsheet_name, fmt='pandas_excel')\nlibraries = book.parse('Libraries', header=0)\ncreated = server.post_sheet('/libraries/', libraries, verbose=True, dry_run=True, validator=validator)\nprint(len(created))", "_____no_output_____" ], [ "if created:\n libraries.to_excel('/dev/shm/libraries.xlsx', index=False)", "_____no_output_____" ] ], [ [ "# Register Experiments", "_____no_output_____" ] ], [ [ "print(server.server)\nbook = gcat.get_file(spreadsheet_name, fmt='pandas_excel')\nexperiments = book.parse('Experiments', header=0)\ncreated = server.post_sheet('/experiments/', experiments, verbose=True, dry_run=False, validator=validator)\nprint(len(created))", "_____no_output_____" ], [ "if created:\n experiments.to_excel('/dev/shm/experiments.xlsx', index=False)", "_____no_output_____" ] ], [ [ "# Register Replicates", "_____no_output_____" ] ], [ [ "print(server.server)\nprint(spreadsheet_name)\nbook = gcat.get_file(spreadsheet_name, fmt='pandas_excel')\nreplicates = book.parse('Replicates', header=0)\ncreated = server.post_sheet('/replicates/', replicates, verbose=True, dry_run=True, validator=validator)\nprint(len(created))", "_____no_output_____" ], [ "if created:\n replicates.to_excel('/dev/shm/replicates.xlsx', index=False)", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "import pandas as pd \nimport numpy as np", "_____no_output_____" ], [ "import matplotlib.pyplot as plt\nimport seaborn as sns", "_____no_output_____" ], [ "%matplotlib inline", "_____no_output_____" ], [ "data = pd.read_csv(\"KNN_Project_Data\")", "_____no_output_____" ], [ "data.info()", "<class 'pandas.core.frame.DataFrame'>\nRangeIndex: 1000 entries, 0 to 999\nData columns (total 11 columns):\nXVPM 1000 non-null float64\nGWYH 1000 non-null float64\nTRAT 1000 non-null float64\nTLLZ 1000 non-null float64\nIGGA 1000 non-null float64\nHYKR 1000 non-null float64\nEDFS 1000 non-null float64\nGUUB 1000 non-null float64\nMGJM 1000 non-null float64\nJHZC 1000 non-null float64\nTARGET CLASS 1000 non-null int64\ndtypes: float64(10), int64(1)\nmemory usage: 86.0 KB\n" ], [ "data.head()", "_____no_output_____" ], [ "from sklearn.preprocessing import StandardScaler", "_____no_output_____" ], [ "scaler = StandardScaler()", "_____no_output_____" ], [ "scaler", "_____no_output_____" ], [ "scaler.fit(data.drop('TARGET CLASS',axis=1))", "_____no_output_____" ], [ "scaled_features = scaler.transform(data.drop('TARGET CLASS',axis=1))", "_____no_output_____" ], [ "df_feat = pd.DataFrame(scaled_features,columns=data.columns[:-1])\ndf_feat.head()", "_____no_output_____" ], [ "from sklearn.model_selection import train_test_split", "_____no_output_____" ], [ "X_train, X_test, y_train, y_test = train_test_split(scaled_features, data['TARGET CLASS'], test_size=0.3, random_state=101)", "_____no_output_____" ], [ "from sklearn.neighbors import KNeighborsClassifier", "_____no_output_____" ], [ "knn = KNeighborsClassifier(n_neighbors=1)", "_____no_output_____" ], [ "knn.fit(X_train,y_train)", "_____no_output_____" ], [ "pred = knn.predict(X_test)", "_____no_output_____" ], [ "from sklearn.metrics import classification_report,confusion_matrix", "_____no_output_____" ], [ "print(confusion_matrix(y_test,pred))", "[[109 43]\n [ 41 107]]\n" ], [ "print(classification_report(y_test,pred))", " precision recall f1-score support\n\n 0 0.73 0.72 0.72 152\n 1 0.71 0.72 0.72 148\n\navg / total 0.72 0.72 0.72 300\n\n" ], [ "error_rate = []\n\n# Will take some time\nfor i in range(1,40):\n \n knn = KNeighborsClassifier(n_neighbors=i)\n knn.fit(X_train,y_train)\n pred_i = knn.predict(X_test)\n error_rate.append(np.mean(pred_i != y_test))", "_____no_output_____" ], [ "plt.figure(figsize=(10,6))\nplt.plot(range(1,40),error_rate,color='blue', linestyle='dashed', marker='o',\n markerfacecolor='red', markersize=10)\nplt.title('Error Rate vs. K Value')\nplt.xlabel('K')\nplt.ylabel('Error Rate')", "_____no_output_____" ], [ "knn = KNeighborsClassifier(n_neighbors=30)\n\nknn.fit(X_train,y_train)\npred = knn.predict(X_test)\n\nprint('WITH K=1')\nprint('\\n')\nprint(confusion_matrix(y_test,pred))\nprint('\\n')\nprint(classification_report(y_test,pred))", "WITH K=1\n\n\n[[124 28]\n [ 24 124]]\n\n\n precision recall f1-score support\n\n 0 0.84 0.82 0.83 152\n 1 0.82 0.84 0.83 148\n\navg / total 0.83 0.83 0.83 300\n\n" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "import numpy as np\nimport pandas as pd\nimport os\nimport scipy\nimport scipy.io\n\nimport sys\n\nimport pystan\nimport matplotlib\nimport matplotlib.pyplot as plt\n\nimport seaborn as sns\n\nnp.random.seed(1234)\n%matplotlib inline", "_____no_output_____" ], [ "patient = 'la01_ictal'\nresultsdir = os.path.join('/Volumes/ADAM LI/pydata/output/pert/', patient)\nfilename = patient + '_pertmodel.npz'\nfiletoload = os.path.join(resultsdir, filename)\n\n# load data\ndata = np.load(filetoload)\npertmats = data['pertmats']\n\nprint pertmats.shape\nprint data.keys()", "/Volumes/ADAM LI/pydata/output/pert/la01_ictal/la01_ictal_pertmodel.npz\n/Volumes/ADAM LI/pydata/converted/\n" ], [ " ### Set the font dictionaries (for plot title and axis titles)\ntitle_font = {'fontname':'Arial', 'size':'34', 'color':'black', 'weight':'normal',\n 'verticalalignment':'bottom'} # Bottom vertical alignment for more space\naxis_font = {'fontname':'Arial', 'size':'30'}\n\n# Begin Plotting\nsns.set(font_scale=2.5)\nfig, ax = plt.subplots(figsize=(30,23))\nsns.heatmap(ax=ax, data=pertmats, cmap=plt.cm.jet, cbar=True, \n cbar_kws={'label': 'Fragility Metric'})\nax.set_title(patient + ' Fragility Heatmap', **title_font)\n", "_____no_output_____" ], [ "perturbation_code = \"\"\"\ndata {\n int<lower=0> N; // number of contacts\n int<lower=0> T; // number of time windows\n real matrix pert[N, T];\n}\nparameters {\n real<lower=0, upper=1> p[N]; // parameters of probability in EZ\n}\nmodel {\n p ~ normal(0.5, 1); // prior on parameter EZ probabilities\n pert ~ normal(0.25, 0.5); // prior on observations perturbation distribution\n \n for (chan in 1:N) {\n fragility[chan] ~ model(p, std)\n }\n}\n\"\"\"\n\n# schools_dat = {'J': 8,\n# 'y': [28, 8, -3, 7, -1, 1, 18, 12],\n# 'sigma': [15, 10, 16, 11, 9, 11, 10, 18]}\n\n# sm = pystan.StanModel(model_code=schools_code)\n# fit = sm.sampling(data=schools_dat, iter=1000, chains=4)\n\nprint fit\nfit.plot()", "Inference for Stan model: anon_model_cbe9cd2f1e5ab5d1c7cce1f23ca970b4.\n4 chains, each with iter=1000; warmup=500; thin=1; \npost-warmup draws per chain=500, total post-warmup draws=2000.\n\n mean se_mean sd 2.5% 25% 50% 75% 97.5% n_eff Rhat\nmu 8.16 0.13 4.78 -0.71 4.96 8.01 11.1 18.29 1294.0 1.0\ntau 6.62 0.17 5.28 0.26 2.53 5.45 9.33 20.18 926.0 1.0\neta[0] 0.38 0.02 0.92 -1.49 -0.22 0.4 1.02 2.12 2000.0 1.0\neta[1] 0.01 0.02 0.87 -1.73 -0.54 7.7e-3 0.58 1.73 2000.0 1.0\neta[2] -0.17 0.02 0.95 -2.08 -0.79 -0.17 0.49 1.67 2000.0 1.0\neta[3] -0.04 0.02 0.86 -1.69 -0.61 -0.05 0.53 1.65 2000.0 1.0\neta[4] -0.38 0.02 0.88 -2.02 -0.96 -0.42 0.16 1.45 2000.0 1.0\neta[5] -0.25 0.02 0.89 -2.02 -0.81 -0.25 0.33 1.51 2000.0 1.0\neta[6] 0.33 0.02 0.87 -1.46 -0.24 0.35 0.91 2.08 2000.0 1.0\neta[7] 0.07 0.02 0.92 -1.74 -0.53 0.07 0.65 1.97 2000.0 1.0\ntheta[0] 11.48 0.21 8.2 -2.08 6.25 10.37 15.87 31.44 1493.0 1.0\ntheta[1] 7.97 0.14 6.29 -4.61 3.97 7.83 11.88 21.02 2000.0 1.0\ntheta[2] 6.47 0.19 7.88 -11.5 2.15 6.86 11.32 21.46 1722.0 1.0\ntheta[3] 7.65 0.14 6.42 -5.85 3.74 7.7 11.71 21.11 2000.0 1.0\ntheta[4] 5.08 0.14 6.33 -8.27 1.31 5.54 9.27 16.44 2000.0 1.0\ntheta[5] 6.11 0.15 6.77 -8.29 2.25 6.52 10.34 18.67 2000.0 1.0\ntheta[6] 10.7 0.15 6.66 -1.37 6.46 10.09 14.68 25.49 2000.0 1.0\ntheta[7] 8.73 0.18 7.89 -6.63 4.18 8.55 12.95 25.0 2000.0 1.0\nlp__ -4.78 0.1 2.71 -10.99 -6.33 -4.54 -2.81 -0.33 668.0 1.0\n\nSamples were drawn using NUTS at Sat Dec 9 14:04:41 2017.\nFor each parameter, n_eff is a crude measure of effective sample size,\nand Rhat is the potential scale reduction factor on split chains (at \nconvergence, Rhat=1).\n" ], [ "la = fit.extract(permuted=True) # return a dictionary of arrays\nmu = la['mu']\n\n## return an array of three dimensions: iterations, chains, parameters\na = fit.extract(permuted=False)", "WARNING:root:`dtypes` ignored when `permuted` is False.\n" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# End-to-end learning for music audio\n\n- http://qiita.com/himono/items/a94969e35fa8d71f876c", "_____no_output_____" ], [ "```\n# データのダウンロード\nwget http://mi.soi.city.ac.uk/datasets/magnatagatune/mp3.zip.001\nwget http://mi.soi.city.ac.uk/datasets/magnatagatune/mp3.zip.002\nwget http://mi.soi.city.ac.uk/datasets/magnatagatune/mp3.zip.003\n\n# 結合\ncat data/mp3.zip.* > data/music.zip\n\n# 解凍\nunzip data/music.zip -d music\n```", "_____no_output_____" ] ], [ [ "%matplotlib inline\nimport os\nimport matplotlib.pyplot as plt", "_____no_output_____" ] ], [ [ "## MP3ファイルのロード", "_____no_output_____" ] ], [ [ "import numpy as np\nfrom pydub import AudioSegment\n\ndef mp3_to_array(file):\n # MP3 => RAW\n song = AudioSegment.from_mp3(file)\n song_arr = np.fromstring(song._data, np.int16)\n return song_arr", "_____no_output_____" ], [ "%ls data/music/1/ambient_teknology-phoenix-01-ambient_teknology-0-29.mp3", "data/music/1/ambient_teknology-phoenix-01-ambient_teknology-0-29.mp3\r\n" ], [ "file = 'data/music/1/ambient_teknology-phoenix-01-ambient_teknology-0-29.mp3'\nsong = mp3_to_array(file)", "_____no_output_____" ], [ "plt.plot(song)", "_____no_output_____" ] ], [ [ "## 楽曲タグデータをロード\n\n- ランダムに3000曲を抽出\n- よく使われるタグ50個を抽出\n- 各曲には複数のタグがついている", "_____no_output_____" ] ], [ [ "import pandas as pd\n\ntags_df = pd.read_csv('data/annotations_final.csv', delim_whitespace=True)\n# 全体をランダムにサンプリング\ntags_df = tags_df.sample(frac=1)\n# 最初の3000曲を使う\ntags_df = tags_df[:3000]\n\ntags_df", "_____no_output_____" ], [ "top50_tags = tags_df.iloc[:, 1:189].sum().sort_values(ascending=False).index[:50].tolist()\ny = tags_df[top50_tags].values\ny", "_____no_output_____" ] ], [ [ "## 楽曲データをロード\n\n- tags_dfのmp3_pathからファイルパスを取得\n- mp3_to_array()でnumpy arrayをロード\n- (samples, features, channels) になるようにreshape\n- 音声波形は1次元なのでchannelsは1\n- 訓練データはすべて同じサイズなのでfeaturesは同じになるはず（パディング不要）", "_____no_output_____" ] ], [ [ "files = tags_df.mp3_path.values\nfiles = [os.path.join('data', 'music', x) for x in files]", "_____no_output_____" ], [ "X = np.array([mp3_to_array(file) for file in files])\nX = X.reshape(X.shape[0], X.shape[1], 1)", "_____no_output_____" ], [ "X.shape", "_____no_output_____" ] ], [ [ "## 訓練データとテストデータに分割", "_____no_output_____" ] ], [ [ "from sklearn.model_selection import train_test_split\nrandom_state = 42\n\ntrain_x, test_x, train_y, test_y = train_test_split(X, y, test_size=0.2, random_state=random_state)", "_____no_output_____" ], [ "print(train_x.shape)\nprint(test_x.shape)\nprint(train_y.shape)\nprint(test_y.shape)", "(2400, 465984, 1)\n(600, 465984, 1)\n(2400, 50)\n(600, 50)\n" ], [ "plt.plot(train_x[0])", "_____no_output_____" ], [ "np.save('train_x.npy', train_x)\nnp.save('test_x.npy', test_x)\nnp.save('train_y.npy', train_y)\nnp.save('test_y.npy', test_y)", "_____no_output_____" ] ], [ [ "## 訓練", "_____no_output_____" ] ], [ [ "import numpy as np\nfrom keras.models import Model\nfrom keras.layers import Dense, Flatten, Input, Conv1D, MaxPooling1D\nfrom keras.callbacks import CSVLogger, ModelCheckpoint\n\ntrain_x = np.load('train_x.npy')\ntrain_y = np.load('train_y.npy')\n\ntest_x = np.load('test_x.npy')\ntest_y = np.load('test_y.npy')\n\nprint(train_x.shape)\nprint(train_y.shape)\nprint(test_x.shape)\nprint(test_y.shape)\n\nfeatures = train_x.shape[1]\n\nx_inputs = Input(shape=(features, 1), name='x_inputs')\nx = Conv1D(128, 256, strides=256, padding='valid', activation='relu')(x_inputs) # strided conv\nx = Conv1D(32, 8, activation='relu')(x)\nx = MaxPooling1D(4)(x)\nx = Conv1D(32, 8, activation='relu')(x)\nx = MaxPooling1D(4)(x)\nx = Conv1D(32, 8, activation='relu')(x)\nx = MaxPooling1D(4)(x)\nx = Conv1D(32, 8, activation='relu')(x)\nx = MaxPooling1D(4)(x)\nx = Flatten()(x)\nx = Dense(100, activation='relu')(x)\nx_outputs = Dense(50, activation='sigmoid', name='x_outputs')(x)\n\nmodel = Model(inputs=x_inputs, outputs=x_outputs)\nmodel.compile(optimizer='adam',\n loss='categorical_crossentropy')\n\nlogger = CSVLogger('history.log')\ncheckpoint = ModelCheckpoint(\n 'model.{epoch:02d}-{val_loss:.3f}.h5',\n monitor='val_loss',\n verbose=1,\n save_best_only=True,\n mode='auto')\n\nmodel.fit(train_x, train_y, batch_size=600, epochs=50,\n validation_data=[test_x, test_y],\n callbacks=[logger, checkpoint])", "_____no_output_____" ] ], [ [ "## 予測\n\n- taggerは複数のタグを出力するのでevaluate()ではダメ？", "_____no_output_____" ] ], [ [ "import numpy as np\nfrom keras.models import load_model\nfrom sklearn.metrics import roc_auc_score\n\ntest_x = np.load('test_x.npy')\ntest_y = np.load('test_y.npy')\n\nmodel = load_model('model.22-9.187-0.202.h5')\n\npred_y = model.predict(test_x, batch_size=50)\nprint(roc_auc_score(test_y, pred_y))\nprint(model.evaluate(test_x, test_y))", "Using TensorFlow backend.\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import pandas as pd\nimport numpy as np\n\nimport requests\n\nimport tensorflow as tf\nimport autokeras as ak\nimport kerastuner\nimport tensorflow_addons as tfa\n\nRS = 69420", "_____no_output_____" ], [ "# Data Download (may take a few minutes depending on your network)\ntrain_datalink_X = 'https://tournament.datacrunch.com/data/X_train.csv' \ntrain_datalink_y = 'https://tournament.datacrunch.com/data/y_train.csv' \nhackathon_data_link = 'https://tournament.datacrunch.com/data/X_test.csv' ", "_____no_output_____" ], [ "# Data for training\ntrain_data = pd.read_csv(train_datalink_X)\n# Data for which you will submit your prediction\ntest_data = pd.read_csv(hackathon_data_link)\n# Targets to be predicted\ntrain_targets = pd.read_csv(train_datalink_y)", "_____no_output_____" ], [ "train_data", "_____no_output_____" ], [ "#If you don't want to work with time series (Later going to produce new models for each moon)\ntrain_data = train_data.drop(columns=['Moons', 'id'])\ntest_data = test_data.drop(columns=['Moons', 'id'])", "_____no_output_____" ], [ "X, y = train_data, train_targets\n\nX.shape, y.shape", "_____no_output_____" ], [ "from sklearn.model_selection import TimeSeriesSplit\ntscv = TimeSeriesSplit()\nprint(tscv)\n\nfor train_index, test_index in tscv.split(X):\n print(\"TRAIN:\", train_index, \"TEST:\", test_index)\n X_train, X_test = X.iloc[train_index], X.iloc[test_index]\n y_train, y_test = y.iloc[train_index], y.iloc[test_index]", "TimeSeriesSplit(max_train_size=None, n_splits=5)\nTRAIN: [ 0 1 2 ... 15177 15178 15179] TEST: [15180 15181 15182 ... 30356 30357 30358]\nTRAIN: [ 0 1 2 ... 30356 30357 30358] TEST: [30359 30360 30361 ... 45535 45536 45537]\nTRAIN: [ 0 1 2 ... 45535 45536 45537] TEST: [45538 45539 45540 ... 60714 60715 60716]\nTRAIN: [ 0 1 2 ... 60714 60715 60716] TEST: [60717 60718 60719 ... 75893 75894 75895]\nTRAIN: [ 0 1 2 ... 75893 75894 75895] TEST: [75896 75897 75898 ... 91072 91073 91074]\n" ], [ "X_train.shape, y_train.shape, y_test.shape, X_test.shape", "_____no_output_____" ], [ "from sklearn.preprocessing import MinMaxScaler\nsc = MinMaxScaler()\n\nX_train = sc.fit_transform(X_train)\nX_test = sc.transform(X_test)", "_____no_output_____" ], [ "# Initialize the structured data classifier.\nrmse_metrics = tf.metrics.RootMeanSquaredError()\nclf = ak.StructuredDataRegressor(overwrite=False,\n project_name='DataCrunchAK',\n directory=r\"D:\\DataCrunch\",\n metrics=[rmse_metrics],\n objective=kerastuner.Objective(\"val_loss\", direction='min'),\n seed=RS,\n max_trials=100)", "INFO:tensorflow:Reloading Oracle from existing project D:\\DataCrunch\\DataCrunchAK\\oracle.json\nINFO:tensorflow:Reloading Tuner from D:\\DataCrunch\\DataCrunchAK\\tuner0.json\n" ], [ "from tensorflow.keras.callbacks import EarlyStopping\nes = EarlyStopping(monitor='val_loss', patience=3, restore_best_weights=True)", "_____no_output_____" ], [ "%%time\nclf.fit(X_train, y_train,\n epochs=75,\n batch_size=512,\n validation_split=0.2,\n callbacks=[es],\n verbose=1)", "Trial 10 Complete [00h 02m 12s]\nval_loss: 0.09039796143770218\n\nBest val_loss So Far: 0.08909150958061218\nTotal elapsed time: 00h 11m 26s\n\nSearch: Running Trial #11\n\nHyperparameter |Value |Best Value So Far \nstructured_data...|True |True \nstructured_data...|False |False \nstructured_data...|3 |3 \nstructured_data...|32 |32 \nstructured_data...|0 |0 \nstructured_data...|32 |32 \nregression_head...|0 |0 \noptimizer |sgd |adam \nlearning_rate |0.001 |0.001 \nstructured_data...|32 |32 \n\nEpoch 1/75\n120/120 [==============================] - 2s 14ms/step - loss: 0.2303 - root_mean_squared_error: 0.4178 - val_loss: 0.1776 - val_root_mean_squared_error: 0.4215\nEpoch 2/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1846 - root_mean_squared_error: 0.4296 - val_loss: 0.1628 - val_root_mean_squared_error: 0.4035\nEpoch 3/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1700 - root_mean_squared_error: 0.4122 - val_loss: 0.1567 - val_root_mean_squared_error: 0.3959\nEpoch 4/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1634 - root_mean_squared_error: 0.4043 - val_loss: 0.1529 - val_root_mean_squared_error: 0.3910\nEpoch 5/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1592 - root_mean_squared_error: 0.3990 - val_loss: 0.1497 - val_root_mean_squared_error: 0.3870\nEpoch 6/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1559 - root_mean_squared_error: 0.3949 - val_loss: 0.1470 - val_root_mean_squared_error: 0.3834\nEpoch 7/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1531 - root_mean_squared_error: 0.3913 - val_loss: 0.1445 - val_root_mean_squared_error: 0.3802\nEpoch 8/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1506 - root_mean_squared_error: 0.3881 - val_loss: 0.1423 - val_root_mean_squared_error: 0.3773\nEpoch 9/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1484 - root_mean_squared_error: 0.3852 - val_loss: 0.1403 - val_root_mean_squared_error: 0.3746\nEpoch 10/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1464 - root_mean_squared_error: 0.3826 - val_loss: 0.1385 - val_root_mean_squared_error: 0.3722\nEpoch 11/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1445 - root_mean_squared_error: 0.3802 - val_loss: 0.1369 - val_root_mean_squared_error: 0.3700\nEpoch 12/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1429 - root_mean_squared_error: 0.3780 - val_loss: 0.1354 - val_root_mean_squared_error: 0.3679\nEpoch 13/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1414 - root_mean_squared_error: 0.3760 - val_loss: 0.1340 - val_root_mean_squared_error: 0.3660\nEpoch 14/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1400 - root_mean_squared_error: 0.3742 - val_loss: 0.1327 - val_root_mean_squared_error: 0.3643\nEpoch 15/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1387 - root_mean_squared_error: 0.3725 - val_loss: 0.1315 - val_root_mean_squared_error: 0.3626\nEpoch 16/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1376 - root_mean_squared_error: 0.3709 - val_loss: 0.1304 - val_root_mean_squared_error: 0.3611\nEpoch 17/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1365 - root_mean_squared_error: 0.3694 - val_loss: 0.1294 - val_root_mean_squared_error: 0.3597\nEpoch 18/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1355 - root_mean_squared_error: 0.3680 - val_loss: 0.1284 - val_root_mean_squared_error: 0.3584\nEpoch 19/75\n120/120 [==============================] - 1s 12ms/step - loss: 0.1345 - root_mean_squared_error: 0.3668 - val_loss: 0.1275 - val_root_mean_squared_error: 0.3571\nEpoch 20/75\n120/120 [==============================] - 1s 12ms/step - loss: 0.1336 - root_mean_squared_error: 0.3656 - val_loss: 0.1267 - val_root_mean_squared_error: 0.3559\nEpoch 21/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1328 - root_mean_squared_error: 0.3644 - val_loss: 0.1259 - val_root_mean_squared_error: 0.3548\nEpoch 22/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1320 - root_mean_squared_error: 0.3633 - val_loss: 0.1251 - val_root_mean_squared_error: 0.3537\nEpoch 23/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1313 - root_mean_squared_error: 0.3623 - val_loss: 0.1244 - val_root_mean_squared_error: 0.3527\nEpoch 24/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1306 - root_mean_squared_error: 0.3614 - val_loss: 0.1237 - val_root_mean_squared_error: 0.3518\nEpoch 25/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1299 - root_mean_squared_error: 0.3604 - val_loss: 0.1231 - val_root_mean_squared_error: 0.3509\nEpoch 26/75\n120/120 [==============================] - 1s 12ms/step - loss: 0.1293 - root_mean_squared_error: 0.3596 - val_loss: 0.1225 - val_root_mean_squared_error: 0.3500\nEpoch 27/75\n120/120 [==============================] - 1s 12ms/step - loss: 0.1287 - root_mean_squared_error: 0.3587 - val_loss: 0.1219 - val_root_mean_squared_error: 0.3492\nEpoch 28/75\n120/120 [==============================] - 1s 12ms/step - loss: 0.1281 - root_mean_squared_error: 0.3580 - val_loss: 0.1214 - val_root_mean_squared_error: 0.3484\nEpoch 29/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1276 - root_mean_squared_error: 0.3572 - val_loss: 0.1209 - val_root_mean_squared_error: 0.3476\nEpoch 30/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1271 - root_mean_squared_error: 0.3565 - val_loss: 0.1203 - val_root_mean_squared_error: 0.3469\nEpoch 31/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1266 - root_mean_squared_error: 0.3558 - val_loss: 0.1199 - val_root_mean_squared_error: 0.3462\nEpoch 32/75\n120/120 [==============================] - 1s 12ms/step - loss: 0.1261 - root_mean_squared_error: 0.3551 - val_loss: 0.1194 - val_root_mean_squared_error: 0.3456\nEpoch 33/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1257 - root_mean_squared_error: 0.3545 - val_loss: 0.1190 - val_root_mean_squared_error: 0.3449\nEpoch 34/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1253 - root_mean_squared_error: 0.3539 - val_loss: 0.1185 - val_root_mean_squared_error: 0.3443\nEpoch 35/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1248 - root_mean_squared_error: 0.3533 - val_loss: 0.1181 - val_root_mean_squared_error: 0.3437\nEpoch 36/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1244 - root_mean_squared_error: 0.3527 - val_loss: 0.1177 - val_root_mean_squared_error: 0.3431\nEpoch 37/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1241 - root_mean_squared_error: 0.3522 - val_loss: 0.1174 - val_root_mean_squared_error: 0.3426\nEpoch 38/75\n120/120 [==============================] - 1s 12ms/step - loss: 0.1237 - root_mean_squared_error: 0.3517 - val_loss: 0.1170 - val_root_mean_squared_error: 0.3421\nEpoch 39/75\n120/120 [==============================] - 1s 12ms/step - loss: 0.1233 - root_mean_squared_error: 0.3512 - val_loss: 0.1167 - val_root_mean_squared_error: 0.3415\nEpoch 40/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1230 - root_mean_squared_error: 0.3507 - val_loss: 0.1163 - val_root_mean_squared_error: 0.3410\nEpoch 41/75\n120/120 [==============================] - 1s 11ms/step - loss: 0.1227 - root_mean_squared_error: 0.3502 - val_loss: 0.1160 - val_root_mean_squared_error: 0.3406\nEpoch 42/75\n120/120 [==============================] - 1s 12ms/step - loss: 0.1224 - root_mean_squared_error: 0.3498 - val_loss: 0.1157 - val_root_mean_squared_error: 0.3401\nEpoch 43/75\n" ], [ "model = clf.export_model()\nmodel.summary()", "_____no_output_____" ], [ "model.evaluate(X_test, y_test)", "_____no_output_____" ], [ "preds = model.predict(test_data)", "_____no_output_____" ], [ "train_targets", "_____no_output_____" ], [ "sc_preds = MinMaxScaler(feature_range=(0.001, 0.999))\npreds_sc = sc_preds.fit_transform(preds)", "_____no_output_____" ], [ "prediction = pd.DataFrame(columns=['target_r', 'target_g', 'target_b'])\nprediction['target_r'] = preds_sc[:, 0]\nprediction['target_g'] = preds_sc[:, 1]\nprediction['target_b'] = preds_sc[:, 2]", "_____no_output_____" ], [ "prediction", "_____no_output_____" ], [ "# prediction.to_csv('1051.csv')", "_____no_output_____" ], [ "prediction.plot(kind='box')", "_____no_output_____" ], [ "API_KEY = 'F1awQwoH9yW7AgG18Nf8XlZNW9xp23b8vY2hHgMxdDimd3u7Z6Q5brcLydHR'\n\nr = requests.post(\"https://tournament.datacrunch.com/api/submission\",\n files = {\n \"file\": (\"x\", prediction.to_csv().encode('ascii'))\n },\n data = {\n \"apiKey\": API_KEY\n },\n)\n\nif r.status_code == 200:\n print(\"Submission submitted :)\")\nelif r.status_code == 423:\n print(\"ERR: Submissions are close\")\n print(\"You can only submit during rounds eg: Friday 7pm GMT+1 to Sunday midnight GMT+1.\")\n print(\"Or the server is currently crunching the submitted files, please wait some time before retrying.\")\nelif r.status_code == 422:\n print(\"ERR: API Key is missing or empty\")\n print(\"Did you forget to fill the API_KEY variable?\")\nelif r.status_code == 404:\n print(\"ERR: Unknown API Key\")\n print(\"You should check that the provided API key is valid and is the same as the one you've received by email.\")\nelif r.status_code == 400:\n print(\"ERR: The file must not be empty\")\n print(\"You have send a empty file.\")\nelif r.status_code == 401:\n print(\"ERR: Your email hasn't been verified\")\n print(\"Please verify your email or contact a cruncher.\")\nelif r.status_code == 429:\n print(\"ERR: Too many submissions\")\nelse:\n print(\"ERR: Server returned: \" + str(r.status_code))\n print(\"Ouch! It seems that we were not expecting this kind of result from the server, if the probleme persist, contact a cruncher.\")", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import numpy as np\nimport pandas as pd\nimport seaborn as sns\nimport matplotlib.pyplot as plt", "_____no_output_____" ], [ "df1 = pd.read_csv('data/Salary_Data.csv')", "_____no_output_____" ], [ "df1.head()", "_____no_output_____" ], [ "df1.shape", "_____no_output_____" ], [ "sns.scatterplot(data=df1,x='YearsExperience',y='Salary')\nplt.show()", "_____no_output_____" ], [ "y = df1.iloc[:,1].values\ny", "_____no_output_____" ], [ "x = df1.iloc[:,:1].values\nx", "_____no_output_____" ] ], [ [ "# Splitting data for Training and Testing", "_____no_output_____" ] ], [ [ "from sklearn.model_selection import train_test_split", "_____no_output_____" ], [ "X_train,X_test,Y_train,Y_test = train_test_split(x,y,train_size=0.7,random_state=0)", "_____no_output_____" ], [ "X_train.shape", "_____no_output_____" ], [ "X_test.shape", "_____no_output_____" ], [ "Y_train.shape", "_____no_output_____" ], [ "Y_test.shape", "_____no_output_____" ] ], [ [ "# Creating a ML Model", "_____no_output_____" ] ], [ [ "from sklearn.linear_model import LinearRegression", "_____no_output_____" ], [ "regressor = LinearRegression()", "_____no_output_____" ], [ "regressor.fit(X_train,Y_train)", "_____no_output_____" ], [ "y_predict = regressor.predict(X_test)", "_____no_output_____" ], [ "y_predict", "_____no_output_____" ], [ "Y_test", "_____no_output_____" ] ], [ [ "# Model Coefficients ", "_____no_output_____" ] ], [ [ "regressor.intercept_", "_____no_output_____" ], [ "regressor.coef_", "_____no_output_____" ] ], [ [ "# Equation of Line --> y = 9360.26* x + 26777.39", "_____no_output_____" ], [ "# Model Evaluation", "_____no_output_____" ] ], [ [ "from sklearn import metrics", "_____no_output_____" ], [ "MAE = metrics.mean_absolute_error(Y_test,y_predict)\nMAE", "_____no_output_____" ], [ "MSE = metrics.mean_squared_error(Y_test,y_predict)\nMSE", "_____no_output_____" ], [ "RMSE = np.sqrt(MSE)\nRMSE", "_____no_output_____" ], [ "R2 = metrics.r2_score(Y_test,y_predict)\nR2", "_____no_output_____" ] ], [ [ "# Hands on Task", "_____no_output_____" ] ], [ [ "df1 = pd.read_csv('data/auto-mpg.csv')\ndf1.head()", "_____no_output_____" ] ], [ [ "# HandsOn - TASK\n\n### 1. Create a Machine Learning Model for predicting Mpg from all the input features in the data.\n\n1. EDA\n2. Data Clearning & Pre-Processing\n3. Build ML Model\n4. Model Evaluation.\n", "_____no_output_____" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "%matplotlib inline", "_____no_output_____" ] ], [ [ "\n# Markers for watershed transform\n\nThe watershed is a classical algorithm used for **segmentation**, that\nis, for separating different objects in an image.\n\nHere a marker image is built from the region of low gradient inside the image.\nIn a gradient image, the areas of high values provide barriers that help to\nsegment the image.\nUsing markers on the lower values will ensure that the segmented objects are\nfound.\n\nSee Wikipedia_ for more details on the algorithm.\n\n", "_____no_output_____" ] ], [ [ "from scipy import ndimage as ndi\nimport matplotlib.pyplot as plt\n\nfrom skimage.morphology import disk\nfrom skimage.segmentation import watershed\nfrom skimage import data\nfrom skimage.filters import rank\nfrom skimage.util import img_as_ubyte\n\n\nimage = img_as_ubyte(data.camera())\n\n# denoise image\ndenoised = rank.median(image, disk(2))\n\n# find continuous region (low gradient -\n# where less than 10 for this image) --> markers\n# disk(5) is used here to get a more smooth image\nmarkers = rank.gradient(denoised, disk(5)) < 10\nmarkers = ndi.label(markers)[0]\n\n# local gradient (disk(2) is used to keep edges thin)\ngradient = rank.gradient(denoised, disk(2))\n\n# process the watershed\nlabels = watershed(gradient, markers)\n\n# display results\nfig, axes = plt.subplots(nrows=2, ncols=2, figsize=(8, 8),\n sharex=True, sharey=True)\nax = axes.ravel()\n\nax[0].imshow(image, cmap=plt.cm.gray)\nax[0].set_title(\"Original\")\n\nax[1].imshow(gradient, cmap=plt.cm.nipy_spectral)\nax[1].set_title(\"Local Gradient\")\n\nax[2].imshow(markers, cmap=plt.cm.nipy_spectral)\nax[2].set_title(\"Markers\")\n\nax[3].imshow(image, cmap=plt.cm.gray)\nax[3].imshow(labels, cmap=plt.cm.nipy_spectral, alpha=.7)\nax[3].set_title(\"Segmented\")\n\nfor a in ax:\n a.axis('off')\n\nfig.tight_layout()\nplt.show()", "_____no_output_____" ] ] ]

[ "code", "markdown", "code" ]

[ [ "code" ], [ "markdown" ], [ "code" ] ]

[ "Apache-2.0", "CC-BY-4.0" ]

[ "Apache-2.0", "CC-BY-4.0" ]

[ "Apache-2.0", "CC-BY-4.0" ]

[ [ [ "<table> <tr>\n <td style=\"background-color:#ffffff;\">\n <a href=\"http://qworld.lu.lv\" target=\"_blank\"><img src=\"..\\images\\qworld.jpg\" width=\"25%\" align=\"left\"> </a></td>\n <td style=\"background-color:#ffffff;vertical-align:bottom;text-align:right;\">\n prepared by <a href=\"http://abu.lu.lv\" target=\"_blank\">Abuzer Yakaryilmaz</a> (<a href=\"http://qworld.lu.lv/index.php/qlatvia/\" target=\"_blank\">QLatvia</a>)\n </td> \n</tr></table>", "_____no_output_____" ], [ "<table width=\"100%\"><tr><td style=\"color:#bbbbbb;background-color:#ffffff;font-size:11px;font-style:italic;text-align:right;\">This cell contains some macros. If there is a problem with displaying mathematical formulas, please run this cell to load these macros. </td></tr></table>\n$ \\newcommand{\\bra}[1]{\\langle #1|} $\n$ \\newcommand{\\ket}[1]{|#1\\rangle} $\n$ \\newcommand{\\braket}[2]{\\langle #1|#2\\rangle} $\n$ \\newcommand{\\dot}[2]{ #1 \\cdot #2} $\n$ \\newcommand{\\biginner}[2]{\\left\\langle #1,#2\\right\\rangle} $\n$ \\newcommand{\\mymatrix}[2]{\\left( \\begin{array}{#1} #2\\end{array} \\right)} $\n$ \\newcommand{\\myvector}[1]{\\mymatrix{c}{#1}} $\n$ \\newcommand{\\myrvector}[1]{\\mymatrix{r}{#1}} $\n$ \\newcommand{\\mypar}[1]{\\left( #1 \\right)} $\n$ \\newcommand{\\mybigpar}[1]{ \\Big( #1 \\Big)} $\n$ \\newcommand{\\sqrttwo}{\\frac{1}{\\sqrt{2}}} $\n$ \\newcommand{\\dsqrttwo}{\\dfrac{1}{\\sqrt{2}}} $\n$ \\newcommand{\\onehalf}{\\frac{1}{2}} $\n$ \\newcommand{\\donehalf}{\\dfrac{1}{2}} $\n$ \\newcommand{\\hadamard}{ \\mymatrix{rr}{ \\sqrttwo & \\sqrttwo \\\\ \\sqrttwo & -\\sqrttwo }} $\n$ \\newcommand{\\vzero}{\\myvector{1\\\\0}} $\n$ \\newcommand{\\vone}{\\myvector{0\\\\1}} $\n$ \\newcommand{\\stateplus}{\\myvector{ \\sqrttwo \\\\ \\sqrttwo } } $\n$ \\newcommand{\\stateminus}{ \\myrvector{ \\sqrttwo \\\\ -\\sqrttwo } } $\n$ \\newcommand{\\myarray}[2]{ \\begin{array}{#1}#2\\end{array}} $\n$ \\newcommand{\\X}{ \\mymatrix{cc}{0 & 1 \\\\ 1 & 0} } $\n$ \\newcommand{\\I}{ \\mymatrix{rr}{1 & 0 \\\\ 0 & 1} } $\n$ \\newcommand{\\Z}{ \\mymatrix{rr}{1 & 0 \\\\ 0 & -1} } $\n$ \\newcommand{\\Htwo}{ \\mymatrix{rrrr}{ \\frac{1}{2} & \\frac{1}{2} & \\frac{1}{2} & \\frac{1}{2} \\\\ \\frac{1}{2} & -\\frac{1}{2} & \\frac{1}{2} & -\\frac{1}{2} \\\\ \\frac{1}{2} & \\frac{1}{2} & -\\frac{1}{2} & -\\frac{1}{2} \\\\ \\frac{1}{2} & -\\frac{1}{2} & -\\frac{1}{2} & \\frac{1}{2} } } $\n$ \\newcommand{\\CNOT}{ \\mymatrix{cccc}{1 & 0 & 0 & 0 \\\\ 0 & 1 & 0 & 0 \\\\ 0 & 0 & 0 & 1 \\\\ 0 & 0 & 1 & 0} } $\n$ \\newcommand{\\norm}[1]{ \\left\\lVert #1 \\right\\rVert } $\n$ \\newcommand{\\pstate}[1]{ \\lceil \\mspace{-1mu} #1 \\mspace{-1.5mu} \\rfloor } $", "_____no_output_____" ], [ "<h2> <font color=\"blue\"> Solutions for </font>Probabilistic Bit</h2>", "_____no_output_____" ], [ "<a id=\"task2\"></a>\n<h3> Task 2 </h3>\n\nSuppose that Fyodor hiddenly rolls a loaded (tricky) dice with the bias \n\n$$ Pr(1):Pr(2):Pr(3):Pr(4):Pr(5):Pr(6) = 7:5:4:2:6:1 . $$\n\nRepresent your information on the result as a column vector. Remark that the size of your column should be 6.\n\nYou may use python for your calculations.", "_____no_output_____" ], [ "<h3>Solution</h3>", "_____no_output_____" ] ], [ [ "# all portions are stored in a list \nall_portions = [7,5,4,2,6,1];\n\n# let's calculate the total portion\ntotal_portion = 0\nfor i in range(6):\n total_portion = total_portion + all_portions[i]\n\nprint(\"total portion is\",total_portion)\n\n# find the weight of one portion\none_portion = 1/total_portion\nprint(\"the weight of one portion is\",one_portion)\n\nprint() # print an empty line\n# now we can calculate the probabilities of rolling 1,2,3,4,5, and 6\nfor i in range(6):\n print(\"the probability of rolling\",(i+1),\"is\",(one_portion*all_portions[i]))", "_____no_output_____" ] ] ]

[ "markdown", "code" ]

[ [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "#@title Paste the values then click run\n#@markdown <a href=\"https://techtanic.github.io/duce\" target=\"_blank\">Website</a>\nemail = \"rfrf445fr@gmail.com\" #@param {type: \"string\"}\npassword = \"1234test\" #@param {type: \"string\"}\n\nimport os\nfor index,item in enumerate([\"requests\",\"bs4\",\"html5lib\",\"colorama\",\"tqdm\",\"cloudscraper\"]):\n print(f\"installing {index}/5\")\n os.system(f\"pip3 install {item} -U\")\n\n\nfrom functools import partial\nfrom tqdm import tqdm\ntqdm = partial(tqdm, position=0, leave=True)\n\nfrom colorama import Fore, Back, Style\n\n# colors foreground text:\nfc = Fore.CYAN\nfg = Fore.GREEN\nfw = Fore.WHITE\nfr = Fore.RED\nfb = Fore.BLUE\nflb = Fore.LIGHTBLUE_EX\nfbl = Fore.BLACK\nfy = Fore.YELLOW\nfm = Fore.MAGENTA\n\n\n# colors background text:\nbc = Back.CYAN\nbg = Back.GREEN\nbw = Back.WHITE\nbr = Back.RED\nbb = Back.BLUE\nby = Back.YELLOW\nbm = Back.MAGENTA\n\n# colors style text:\nsd = Style.DIM\nsn = Style.NORMAL\nsb = Style.BRIGHT\n\n\nimport json\nimport random\nimport re\nimport threading\nimport time\nimport traceback\nfrom urllib.parse import parse_qs, unquote, urlsplit\nfrom decimal import Decimal\nimport requests\nimport cloudscraper\nfrom bs4 import BeautifulSoup as bs\n\n# DUCE-CLI\ndef remove_suffix(input_string:str, suffix:str)->str:\n if suffix and input_string.endswith(suffix):\n return input_string[: -len(suffix)]\n return input_string\n\n\ndef remove_prefix(input_string:str, prefix:str)->str:\n if prefix and input_string.startswith(prefix):\n return input_string[len(prefix) :]\n return input_string\n# Scraper\ndef discudemy():\n global du_links\n du_links = []\n big_all = []\n head = {\n \"User-Agent\": \"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/89.0.4389.128 Safari/537.36 Edg/89.0.774.77\",\n \"Accept\": \"text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.9\",\n }\n\n for page in range(1, 4):\n r = requests.get(\"https://www.discudemy.com/all/\" + str(page), headers=head)\n soup = bs(r.content, \"html5lib\")\n small_all = soup.find_all(\"section\", \"card\")\n big_all.extend(small_all)\n du_bar = tqdm(total=len(big_all), desc=\"Discudemy\")\n for index, item in enumerate(big_all):\n du_bar.update(1)\n title = item.string\n url = item[\"href\"].split(\"/\")[4]\n r = requests.get(\"https://www.discudemy.com/go/\" + url, headers=head)\n soup = bs(r.content, \"html5lib\")\n du_links.append(title + \"|:|\" + soup.find(\"a\", id=\"couponLink\").string)\n du_bar.close()\n\n\ndef udemy_freebies():\n global uf_links\n uf_links = []\n big_all = []\n\n for page in range(1, 3):\n r = requests.get(\n \"https://www.udemyfreebies.com/free-udemy-courses/\" + str(page)\n )\n soup = bs(r.content, \"html5lib\")\n small_all = soup.find_all(\"a\", {\"class\": \"theme-img\"})\n big_all.extend(small_all)\n uf_bar = tqdm(total=len(big_all), desc=\"Udemy Freebies\")\n\n for index, item in enumerate(big_all):\n uf_bar.update(1)\n title = item.img[\"alt\"]\n link = requests.get(\n \"https://www.udemyfreebies.com/out/\" + item[\"href\"].split(\"/\")[4]\n ).url\n uf_links.append(title + \"|:|\" + link)\n uf_bar.close()\n\n\ndef tutorialbar():\n\n global tb_links\n tb_links = []\n big_all = []\n\n for page in range(1, 4):\n r = requests.get(\"https://www.tutorialbar.com/all-courses/page/\" + str(page))\n soup = bs(r.content, \"html5lib\")\n small_all = soup.find_all(\n \"h3\", class_=\"mb15 mt0 font110 mobfont100 fontnormal lineheight20\"\n )\n big_all.extend(small_all)\n tb_bar = tqdm(total=len(big_all), desc=\"Tutorial Bar\")\n\n for index, item in enumerate(big_all):\n tb_bar.update(1)\n title = item.a.string\n url = item.a[\"href\"]\n r = requests.get(url)\n soup = bs(r.content, \"html5lib\")\n link = soup.find(\"a\", class_=\"btn_offer_block re_track_btn\")[\"href\"]\n if \"www.udemy.com\" in link:\n tb_links.append(title + \"|:|\" + link)\n tb_bar.close()\n\n\ndef real_discount():\n\n global rd_links\n rd_links = []\n big_all = []\n\n for page in range(1, 3):\n r = requests.get(\"https://real.discount/stores/Udemy?page=\" + str(page))\n soup = bs(r.content, \"html5lib\")\n small_all = soup.find_all(\"div\", class_=\"col-xl-4 col-md-6\")\n big_all.extend(small_all)\n rd_bar = tqdm(total=len(big_all), desc=\"Real Discount\")\n\n for index, item in enumerate(big_all):\n rd_bar.update(1)\n title = item.h3.string\n url = \"https://real.discount\" + item.a[\"href\"]\n r = requests.get(url)\n soup = bs(r.content, \"html5lib\")\n link = soup.find(\"div\", class_=\"col-xs-12 col-md-12 col-sm-12 text-center\").a[\n \"href\"\n ]\n if link.startswith(\"http://click.linksynergy.com\"):\n link = parse_qs(link)[\"RD_PARM1\"][0]\n rd_links.append(title+\"|:|\"+link)\n rd_bar.close()\n\n\ndef coursevania():\n\n global cv_links\n cv_links = []\n r = requests.get(\"https://coursevania.com/courses/\")\n soup = bs(r.content, \"html5lib\")\n\n nonce = json.loads([script.string for script in soup.find_all('script') if script.string and \"load_content\" in script.string][0].strip(\"_mlv = norsecat;\\n\"))[\"load_content\"]\n\n r = requests.get(\n \"https://coursevania.com/wp-admin/admin-ajax.php?&template=courses/grid&args={%22posts_per_page%22:%2230%22}&action=stm_lms_load_content&nonce=\"\n + nonce\n + \"&sort=date_high\"\n ).json()\n soup = bs(r[\"content\"], \"html5lib\")\n small_all = soup.find_all(\"div\", {\"class\": \"stm_lms_courses__single--title\"})\n cv_bar = tqdm(total=len(small_all), desc=\"Course Vania\")\n\n for index, item in enumerate(small_all):\n cv_bar.update(1)\n title = item.h5.string\n r = requests.get(item.a[\"href\"])\n soup = bs(r.content, \"html5lib\")\n cv_links.append(\n title + \"|:|\" + soup.find(\"div\", {\"class\": \"stm-lms-buy-buttons\"}).a[\"href\"]\n )\n cv_bar.close()\n\n\ndef idcoupons():\n\n global idc_links\n idc_links = []\n big_all = []\n for page in range(1, 6):\n r = requests.get(\n \"https://idownloadcoupon.com/product-category/udemy-2/page/\" + str(page)\n )\n soup = bs(r.content, \"html5lib\")\n small_all = soup.find_all(\"a\", attrs={\"class\": \"button product_type_external\"})\n big_all.extend(small_all)\n idc_bar = tqdm(total=len(big_all), desc=\"IDownloadCoupons\")\n\n for index, item in enumerate(big_all):\n idc_bar.update(1)\n title = item[\"aria-label\"]\n link = unquote(item[\"href\"])\n if link.startswith(\"https://ad.admitad.com\"):\n link = parse_qs(link)[\"ulp\"][0]\n elif link.startswith(\"https://click.linksynergy.com\"):\n link = parse_qs(link)[\"murl\"][0]\n idc_links.append(title + \"|:|\" + link)\n idc_bar.close()\n\ndef enext() -> list:\n en_links = []\n r = requests.get(\"https://e-next.in/e/udemycoupons.php\")\n soup = bs(r.content, \"html5lib\")\n big_all = soup.find(\"div\", {\"class\": \"scroll-box\"}).find_all(\"p\", {\"class\": \"p2\"})\n en_bar = tqdm(total=len(big_all), desc=\"E-next\")\n for i in big_all:\n en_bar.update(1)\n title = i.text[11:].strip().removesuffix(\"Enroll Now free\").strip()\n link = i.a[\"href\"]\n en_links.append(title + \"|:|\" + link)\n en_bar.close()\n\n\n# Constants\n\nversion = \"v1.6\"\n\n\ndef create_scrape_obj():\n funcs = {\n \"Discudemy\": threading.Thread(target=discudemy, daemon=True),\n \"Udemy Freebies\": threading.Thread(target=udemy_freebies, daemon=True),\n \"Tutorial Bar\": threading.Thread(target=tutorialbar, daemon=True),\n \"Real Discount\": threading.Thread(target=real_discount, daemon=True),\n \"Course Vania\": threading.Thread(target=coursevania, daemon=True),\n \"IDownloadCoupons\": threading.Thread(target=idcoupons, daemon=True),\n \"E-next\": threading.Thread(target=enext, daemon=True),\n }\n return funcs\n\n\n################\n\n\n\ndef cookiejar(\n client_id,\n access_token,\n csrf_token,\n):\n cookies = dict(\n client_id=client_id,\n access_token=access_token,\n csrf_token=csrf_token,\n )\n return cookies\n\ndef get_course_id(url):\n r = requests.get(url, allow_redirects=False)\n if r.status_code in (404, 302, 301):\n return False\n if \"/course/draft/\" in url:\n return False\n soup = bs(r.content, \"html5lib\")\n\n try:\n courseid = soup.find(\n \"div\",\n attrs={\"data-content-group\": \"Landing Page\"},\n )[\"data-course-id\"]\n except:\n courseid = soup.find(\n \"body\", attrs={\"data-module-id\": \"course-landing-page/udlite\"}\n )[\"data-clp-course-id\"]\n # with open(\"problem.txt\",\"w\",encoding=\"utf-8\") as f:\n # f.write(str(soup))\n return courseid\n\n\ndef get_course_coupon(url):\n query = urlsplit(url).query\n params = parse_qs(query)\n try:\n params = {k: v[0] for k, v in params.items()}\n return params[\"couponCode\"]\n except:\n return \"\"\n\n\n\ndef course_landing_api(courseid):\n r = s.get(\n \"https://www.udemy.com/api-2.0/course-landing-components/\"\n + courseid\n + \"/me/?components=purchase\"\n ).json()\n try:\n purchased = r[\"purchase\"][\"data\"][\"purchase_date\"]\n except:\n purchased = False\n try:\n amount = r[\"purchase\"][\"data\"][\"list_price\"][\"amount\"]\n except:\n print(r[\"purchase\"][\"data\"])\n\n return purchased, Decimal(amount)\n\n\ndef remove_duplicates(l):\n l = l[::-1]\n for i in l:\n while l.count(i) > 1:\n l.remove(i)\n return l[::-1]\n\ndef update_available():\n if remove_prefix(version,\"v\") < remove_prefix(requests.get(\n \"https://api.github.com/repos/techtanic/Discounted-Udemy-Course-Enroller/releases/latest\"\n ).json()[\"tag_name\"],\"v\"):\n print(by + fr + \" Update Available \")\n else:\n return\n\n\ndef check_login():\n \n for retry in range(4):\n \n s = cloudscraper.CloudScraper()\n\n r = s.get(\n \"https://www.udemy.com/join/signup-popup/\",\n )\n soup = bs(r.text, \"html5lib\")\n\n csrf_token = soup.find(\"input\", {\"name\": \"csrfmiddlewaretoken\"})[\"value\"]\n\n data = {\n \"email\": email,\n \"password\": password,\n \"locale\": \"en_US\",\n \"csrfmiddlewaretoken\": csrf_token,\n }\n \n s.headers.update({\"Referer\": \"https://www.udemy.com/join/signup-popup/\"})\n try:\n r = s.post(\n \"https://www.udemy.com/join/login-popup/?locale=en_US\",\n data=data,\n allow_redirects=False,\n )\n except cloudscraper.exceptions.CloudflareChallengeError:\n if retry == 3:\n print(\"Cloudflare is blocking your requests try again after an hour\")\n !kill -9 -1\n retry -= 1\n continue\n if r.status_code != 302:\n soup = bs(r.content, \"html5lib\")\n txt = soup.find(\"div\", class_=\"alert alert-danger js-error-alert\").text.strip()\n if txt[0] == \"Y\":\n print(\"Too many logins per hour try later\")\n elif txt[0] == \"T\":\n print(\"Email or password incorrect\")\n else:\n print(txt)\n time.sleep(1)\n !kill -9 -1\n cookies = cookiejar(r.cookies[\"client_id\"], r.cookies[\"access_token\"], csrf_token)\n\n head = {\n \"authorization\": \"Bearer \" + r.cookies[\"access_token\"],\n \"accept\": \"application/json, text/plain, */*\",\n \"x-requested-with\": \"XMLHttpRequest\",\n \"user-agent\": \"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/89.0.4389.128 Safari/537.36 Edg/89.0.774.77\",\n \"x-forwarded-for\": str(\n \".\".join(map(str, (random.randint(0, 255) for _ in range(4))))\n ),\n \"x-udemy-authorization\": \"Bearer \" + r.cookies[\"access_token\"],\n \"content-type\": \"application/json;charset=UTF-8\",\n \"origin\": \"https://www.udemy.com\",\n \"referer\": \"https://www.udemy.com/\",\n \"dnt\": \"1\",\n }\n \n s = requests.session()\n s.cookies.update(cookies)\n s.headers.update(head)\n s.keep_alive = False\n \n r = s.get(\n \"https://www.udemy.com/api-2.0/contexts/me/?me=True&Config=True\"\n ).json()\n currency = r[\"Config\"][\"price_country\"][\"currency\"]\n user = \"\"\n user = r[\"me\"][\"display_name\"]\n\n return head, user, currency, s\n\n\n# -----------------\ndef free_checkout(coupon, courseid):\n payload = (\n '{\"checkout_environment\":\"Marketplace\",\"checkout_event\":\"Submit\",\"shopping_info\":{\"items\":[{\"discountInfo\":{\"code\":\"'\n + coupon\n + '\"},\"buyable\":{\"type\":\"course\",\"id\":'\n + str(courseid)\n + ',\"context\":{}},\"price\":{\"amount\":0,\"currency\":\"'\n + currency\n + '\"}}]},\"payment_info\":{\"payment_vendor\":\"Free\",\"payment_method\":\"free-method\"}}'\n )\n\n r = s.post(\n \"https://www.udemy.com/payment/checkout-submit/\",\n data=payload,\n verify=False,\n )\n return r.json()\n\n\ndef free_enroll(courseid):\n\n s.get(\n \"https://www.udemy.com/course/subscribe/?courseId=\" + str(courseid)\n )\n\n r = s.get(\n \"https://www.udemy.com/api-2.0/users/me/subscribed-courses/\"\n + str(courseid)\n + \"/?fields%5Bcourse%5D=%40default%2Cbuyable_object_type%2Cprimary_subcategory%2Cis_private\"\n )\n return r.json()\n\n\n# -----------------\n\n\ndef auto(list_st):\n\n se_c, ae_c, e_c, ex_c, as_c = 0, 0, 0, 0, 0\n for index, link in enumerate(list_st):\n title = link.split(\"|:|\")\n print(fy + str(index) + \" \" + title[0], end=\" \")\n link = title[1]\n print(fb + link)\n course_id = get_course_id(link)\n if course_id:\n coupon_id = get_course_coupon(link)\n purchased, amount = course_landing_api(course_id)\n\n if not purchased:\n\n if coupon_id:\n slp = \"\"\n\n js = free_checkout(coupon_id, course_id)\n try:\n if js[\"status\"] == \"succeeded\":\n print(fg + \"Successfully Enrolled\\n\")\n se_c += 1\n as_c += amount\n\n elif js[\"status\"] == \"failed\":\n # print(js)\n print(fr + \"Coupon Expired\\n\")\n e_c += 1\n\n except:\n try:\n msg = js[\"detail\"]\n print(fr + msg)\n print()\n slp = int(re.search(r\"\\d+\", msg).group(0))\n except:\n # print(js)\n print(fr + \"Expired Coupon\\n\")\n e_c += 1\n\n if slp != \"\":\n slp += 5\n print(\n fr\n + \">>> Pausing execution of script for \"\n + str(slp)\n + \" seconds\\n\",\n )\n time.sleep(slp)\n else:\n time.sleep(4)\n\n elif not coupon_id:\n js = free_enroll(course_id)\n try:\n if js[\"_class\"] == \"course\":\n print(fg + \"Successfully Subscribed\\n\")\n se_c += 1\n as_c += amount\n\n except:\n print(fr + \"COUPON MIGHT HAVE EXPIRED\\n\")\n e_c += 1\n\n elif purchased:\n print(flb + purchased)\n print()\n ae_c += 1\n\n elif not course_id:\n print(fr + \"Course Doesn't exist\\n\")\n\n print(f\"Successfully Enrolled: {se_c}\")\n print(f\"Already Enrolled: {ae_c}\")\n print(f\"Amount Saved: ${round(as_c,2)}\")\n print(f\"Expired Courses: {e_c}\")\n print(f\"Excluded Courses: {ex_c}\")\n\n\ndef random_color():\n col = [\"green\", \"yellow\", \"white\"]\n return random.choice(col)\n\n\n##########################################\n\n\ndef main1():\n try:\n links_ls = []\n for index in all_functions:\n all_functions[index].start()\n time.sleep(0.09)\n for t in all_functions:\n all_functions[t].join()\n time.sleep(1)\n\n for link_list in [\n \"du_links\",\n \"uf_links\",\n \"tb_links\",\n \"rd_links\",\n \"cv_links\",\n \"idc_links\",\n \"en_links\",\n ]:\n try:\n links_ls += eval(link_list)\n except:\n pass\n\n auto(remove_duplicates(links_ls))\n\n except:\n e = traceback.format_exc()\n print(e)\n\n############## MAIN ############# MAIN############## MAIN ############# MAIN ############## MAIN ############# MAIN ###########\n\nprint(fb+\"Trying to login\")\ntry:\n head, user, currency, s= check_login()\n print(fg+f\"Logged in as {user}\")\nexcept Exception as e:\n print(fr+f\"Login Error\")\n e = traceback.format_exc()\n print(e)\ntry:\n update_available()\nexcept:\n pass\n\n\nall_functions = create_scrape_obj()\ntm = threading.Thread(target=main1, daemon=True)\ntm.start()\n\ntm.join()\ntry:\n update_available()\nexcept:\n pass", "installing 0/5\ninstalling 1/5\ninstalling 2/5\ninstalling 3/5\ninstalling 4/5\ninstalling 5/5\n\u001b[34mTrying to login\nToo many logins per hour try later\n" ] ] ]

[ "code" ]

[ [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "empty" ] ] ]

[ "empty" ]

[ [ "empty" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Porto Seguro's Safe Driving Prediction\n\nPorto Seguro, one of Brazil’s largest auto and homeowner insurance companies, completely agrees. Inaccuracies in car insurance company’s claim predictions raise the cost of insurance for good drivers and reduce the price for bad ones.\n\nIn the [Porto Seguro Safe Driver Prediction competition](https://www.kaggle.com/c/porto-seguro-safe-driver-prediction), the challenge is to build a model that predicts the probability that a driver will initiate an auto insurance claim in the next year. While Porto Seguro has used machine learning for the past 20 years, they’re looking to Kaggle’s machine learning community to explore new, more powerful methods. A more accurate prediction will allow them to further tailor their prices, and hopefully make auto insurance coverage more accessible to more drivers.\n\nLucky for you, a machine learning model was built to solve the Porto Seguro problem by the data scientist on your team. The solution notebook has steps to load data, split the data into test and train sets, train, evaluate and save a LightGBM model that will be used for the future challenges.\n\n#### Hint: use shift + enter to run the code cells below. Once the cell turns from [*] to [#], you can be sure the cell has run. ", "_____no_output_____" ], [ "## Import Needed Packages\n\nImport the packages needed for this solution notebook. The most widely used packages for machine learning for [scikit-learn](https://scikit-learn.org/stable/), [pandas](https://pandas.pydata.org/docs/getting_started/index.html#getting-started), and [numpy](https://numpy.org/). These packages have various features, as well as a lot of clustering, regression and classification algorithms that make it a good choice for data mining and data analysis. In this notebook, we're using a training function from [lightgbm](https://lightgbm.readthedocs.io/en/latest/index.html).", "_____no_output_____" ] ], [ [ "import os\nimport numpy as np\nimport pandas as pd\nimport lightgbm\nfrom sklearn.model_selection import train_test_split\nimport joblib\nfrom sklearn import metrics", "_____no_output_____" ] ], [ [ "## Load Data\n\nLoad the training dataset from the ./data/ directory. Df.shape() allows you to view the dimensions of the dataset you are passing in. If you want to view the first 5 rows of data, df.head() allows for this.", "_____no_output_____" ] ], [ [ "DATA_DIR = \"../data\"\ndata_df = pd.read_csv(os.path.join(DATA_DIR, 'porto_seguro_safe_driver_prediction_input.csv'))\nprint(data_df.shape)\ndata_df.head()", "(595212, 59)\n" ] ], [ [ "## Split Data into Train and Validatation Sets\n\nPartitioning data into training, validation, and holdout sets allows you to develop highly accurate models that are relevant to data that you collect in the future, not just the data the model was trained on. \n\nIn machine learning, features are the measurable property of the object you’re trying to analyze. Typically, features are the columns of the data that you are training your model with minus the label. In machine learning, a label (categorical) or target (regression) is the output you get from your model after training it.", "_____no_output_____" ] ], [ [ "features = data_df.drop(['target', 'id'], axis = 1)\nlabels = np.array(data_df['target'])\nfeatures_train, features_valid, labels_train, labels_valid = train_test_split(features, labels, test_size=0.2, random_state=0)\n\ntrain_data = lightgbm.Dataset(features_train, label=labels_train)\nvalid_data = lightgbm.Dataset(features_valid, label=labels_valid, free_raw_data=False)", "_____no_output_____" ] ], [ [ "## Train Model\n\nA machine learning model is an algorithm which learns features from the given data to produce labels which may be continuous or categorical ( regression and classification respectively ). In other words, it tries to relate the given data with its labels, just as the human brain does.\n\nIn this cell, the data scientist used an algorithm called [LightGBM](https://lightgbm.readthedocs.io/en/latest/), which primarily used for unbalanced datasets. AUC will be explained in the next cell.", "_____no_output_____" ] ], [ [ "parameters = {\n 'learning_rate': 0.02,\n 'boosting_type': 'gbdt',\n 'objective': 'binary',\n 'metric': 'auc',\n 'sub_feature': 0.7,\n 'num_leaves': 60,\n 'min_data': 100,\n 'min_hessian': 1,\n 'verbose': 4\n}\n \nmodel = lightgbm.train(parameters,\n train_data,\n valid_sets=valid_data,\n num_boost_round=500,\n early_stopping_rounds=20)", "[1]\tvalid_0's auc: 0.595844\nTraining until validation scores don't improve for 20 rounds\n[2]\tvalid_0's auc: 0.605252\n[3]\tvalid_0's auc: 0.612784\n[4]\tvalid_0's auc: 0.61756\n[5]\tvalid_0's auc: 0.620129\n[6]\tvalid_0's auc: 0.622447\n[7]\tvalid_0's auc: 0.622163\n[8]\tvalid_0's auc: 0.622112\n[9]\tvalid_0's auc: 0.622581\n[10]\tvalid_0's auc: 0.622278\n[11]\tvalid_0's auc: 0.622433\n[12]\tvalid_0's auc: 0.623423\n[13]\tvalid_0's auc: 0.623618\n[14]\tvalid_0's auc: 0.62414\n[15]\tvalid_0's auc: 0.624421\n[16]\tvalid_0's auc: 0.624512\n[17]\tvalid_0's auc: 0.625151\n[18]\tvalid_0's auc: 0.62529\n[19]\tvalid_0's auc: 0.625437\n[20]\tvalid_0's auc: 0.62563\n[21]\tvalid_0's auc: 0.625963\n[22]\tvalid_0's auc: 0.626147\n[23]\tvalid_0's auc: 0.626383\n[24]\tvalid_0's auc: 0.626618\n[25]\tvalid_0's auc: 0.626586\n[26]\tvalid_0's auc: 0.626839\n[27]\tvalid_0's auc: 0.626972\n[28]\tvalid_0's auc: 0.626935\n[29]\tvalid_0's auc: 0.626946\n[30]\tvalid_0's auc: 0.627204\n[31]\tvalid_0's auc: 0.627252\n[32]\tvalid_0's auc: 0.627302\n[33]\tvalid_0's auc: 0.627249\n[34]\tvalid_0's auc: 0.627517\n[35]\tvalid_0's auc: 0.627755\n[36]\tvalid_0's auc: 0.62766\n[37]\tvalid_0's auc: 0.627483\n[38]\tvalid_0's auc: 0.627578\n[39]\tvalid_0's auc: 0.627433\n[40]\tvalid_0's auc: 0.627573\n[41]\tvalid_0's auc: 0.627908\n[42]\tvalid_0's auc: 0.627968\n[43]\tvalid_0's auc: 0.628082\n[44]\tvalid_0's auc: 0.628398\n[45]\tvalid_0's auc: 0.628763\n[46]\tvalid_0's auc: 0.629011\n[47]\tvalid_0's auc: 0.629321\n[48]\tvalid_0's auc: 0.629341\n[49]\tvalid_0's auc: 0.629353\n[50]\tvalid_0's auc: 0.629291\n[51]\tvalid_0's auc: 0.629447\n[52]\tvalid_0's auc: 0.629507\n[53]\tvalid_0's auc: 0.629725\n[54]\tvalid_0's auc: 0.630048\n[55]\tvalid_0's auc: 0.630085\n[56]\tvalid_0's auc: 0.630035\n[57]\tvalid_0's auc: 0.630236\n[58]\tvalid_0's auc: 0.630486\n[59]\tvalid_0's auc: 0.630663\n[60]\tvalid_0's auc: 0.630787\n[61]\tvalid_0's auc: 0.630932\n[62]\tvalid_0's auc: 0.631004\n[63]\tvalid_0's auc: 0.631161\n[64]\tvalid_0's auc: 0.631407\n[65]\tvalid_0's auc: 0.631408\n[66]\tvalid_0's auc: 0.631515\n[67]\tvalid_0's auc: 0.631631\n[68]\tvalid_0's auc: 0.631628\n[69]\tvalid_0's auc: 0.631703\n[70]\tvalid_0's auc: 0.631781\n[71]\tvalid_0's auc: 0.631786\n[72]\tvalid_0's auc: 0.631779\n[73]\tvalid_0's auc: 0.632022\n[74]\tvalid_0's auc: 0.632086\n[75]\tvalid_0's auc: 0.632107\n[76]\tvalid_0's auc: 0.632201\n[77]\tvalid_0's auc: 0.632165\n[78]\tvalid_0's auc: 0.632335\n[79]\tvalid_0's auc: 0.632446\n[80]\tvalid_0's auc: 0.63254\n[81]\tvalid_0's auc: 0.632654\n[82]\tvalid_0's auc: 0.632663\n[83]\tvalid_0's auc: 0.632811\n[84]\tvalid_0's auc: 0.63291\n[85]\tvalid_0's auc: 0.632993\n[86]\tvalid_0's auc: 0.632962\n[87]\tvalid_0's auc: 0.632941\n[88]\tvalid_0's auc: 0.633062\n[89]\tvalid_0's auc: 0.633144\n[90]\tvalid_0's auc: 0.633242\n[91]\tvalid_0's auc: 0.633336\n[92]\tvalid_0's auc: 0.633453\n[93]\tvalid_0's auc: 0.633556\n[94]\tvalid_0's auc: 0.633648\n[95]\tvalid_0's auc: 0.633762\n[96]\tvalid_0's auc: 0.633831\n[97]\tvalid_0's auc: 0.633922\n[98]\tvalid_0's auc: 0.633908\n[99]\tvalid_0's auc: 0.633958\n[100]\tvalid_0's auc: 0.634122\n[101]\tvalid_0's auc: 0.634278\n[102]\tvalid_0's auc: 0.634301\n[103]\tvalid_0's auc: 0.634313\n[104]\tvalid_0's auc: 0.634366\n[105]\tvalid_0's auc: 0.634497\n[106]\tvalid_0's auc: 0.634442\n[107]\tvalid_0's auc: 0.634487\n[108]\tvalid_0's auc: 0.634578\n[109]\tvalid_0's auc: 0.634676\n[110]\tvalid_0's auc: 0.63479\n[111]\tvalid_0's auc: 0.634846\n[112]\tvalid_0's auc: 0.634918\n[113]\tvalid_0's auc: 0.63501\n[114]\tvalid_0's auc: 0.634965\n[115]\tvalid_0's auc: 0.635029\n[116]\tvalid_0's auc: 0.635077\n[117]\tvalid_0's auc: 0.635075\n[118]\tvalid_0's auc: 0.6352\n[119]\tvalid_0's auc: 0.635215\n[120]\tvalid_0's auc: 0.635231\n[121]\tvalid_0's auc: 0.635276\n[122]\tvalid_0's auc: 0.635268\n[123]\tvalid_0's auc: 0.635221\n[124]\tvalid_0's auc: 0.635178\n[125]\tvalid_0's auc: 0.635221\n[126]\tvalid_0's auc: 0.635288\n[127]\tvalid_0's auc: 0.635345\n[128]\tvalid_0's auc: 0.635348\n[129]\tvalid_0's auc: 0.635414\n[130]\tvalid_0's auc: 0.635418\n[131]\tvalid_0's auc: 0.635352\n[132]\tvalid_0's auc: 0.635402\n[133]\tvalid_0's auc: 0.635497\n[134]\tvalid_0's auc: 0.635545\n[135]\tvalid_0's auc: 0.63565\n[136]\tvalid_0's auc: 0.635622\n[137]\tvalid_0's auc: 0.635664\n[138]\tvalid_0's auc: 0.635781\n[139]\tvalid_0's auc: 0.635735\n[140]\tvalid_0's auc: 0.635719\n[141]\tvalid_0's auc: 0.635815\n[142]\tvalid_0's auc: 0.635799\n[143]\tvalid_0's auc: 0.63583\n[144]\tvalid_0's auc: 0.635898\n[145]\tvalid_0's auc: 0.635924\n[146]\tvalid_0's auc: 0.635885\n[147]\tvalid_0's auc: 0.635919\n[148]\tvalid_0's auc: 0.63598\n[149]\tvalid_0's auc: 0.636035\n[150]\tvalid_0's auc: 0.636087\n[151]\tvalid_0's auc: 0.636139\n[152]\tvalid_0's auc: 0.63617\n[153]\tvalid_0's auc: 0.636128\n[154]\tvalid_0's auc: 0.636096\n[155]\tvalid_0's auc: 0.636206\n[156]\tvalid_0's auc: 0.636259\n[157]\tvalid_0's auc: 0.636289\n[158]\tvalid_0's auc: 0.636283\n[159]\tvalid_0's auc: 0.636287\n[160]\tvalid_0's auc: 0.636293\n[161]\tvalid_0's auc: 0.636324\n[162]\tvalid_0's auc: 0.63633\n[163]\tvalid_0's auc: 0.636367\n[164]\tvalid_0's auc: 0.636438\n[165]\tvalid_0's auc: 0.636483\n[166]\tvalid_0's auc: 0.636577\n[167]\tvalid_0's auc: 0.636645\n[168]\tvalid_0's auc: 0.63659\n[169]\tvalid_0's auc: 0.636595\n[170]\tvalid_0's auc: 0.636672\n[171]\tvalid_0's auc: 0.636719\n[172]\tvalid_0's auc: 0.636755\n[173]\tvalid_0's auc: 0.636833\n[174]\tvalid_0's auc: 0.636908\n[175]\tvalid_0's auc: 0.636929\n[176]\tvalid_0's auc: 0.636928\n[177]\tvalid_0's auc: 0.636962\n[178]\tvalid_0's auc: 0.636969\n[179]\tvalid_0's auc: 0.636995\n[180]\tvalid_0's auc: 0.637059\n[181]\tvalid_0's auc: 0.637089\n[182]\tvalid_0's auc: 0.637085\n[183]\tvalid_0's auc: 0.637121\n[184]\tvalid_0's auc: 0.637131\n[185]\tvalid_0's auc: 0.637133\n[186]\tvalid_0's auc: 0.637144\n[187]\tvalid_0's auc: 0.637189\n[188]\tvalid_0's auc: 0.637173\n[189]\tvalid_0's auc: 0.63719\n[190]\tvalid_0's auc: 0.637205\n[191]\tvalid_0's auc: 0.637131\n[192]\tvalid_0's auc: 0.637159\n[193]\tvalid_0's auc: 0.637185\n[194]\tvalid_0's auc: 0.63719\n[195]\tvalid_0's auc: 0.637224\n[196]\tvalid_0's auc: 0.637219\n[197]\tvalid_0's auc: 0.637193\n[198]\tvalid_0's auc: 0.637297\n[199]\tvalid_0's auc: 0.637329\n[200]\tvalid_0's auc: 0.6373\n[201]\tvalid_0's auc: 0.637257\n[202]\tvalid_0's auc: 0.637253\n[203]\tvalid_0's auc: 0.637261\n[204]\tvalid_0's auc: 0.637252\n[205]\tvalid_0's auc: 0.637273\n[206]\tvalid_0's auc: 0.637297\n[207]\tvalid_0's auc: 0.637345\n[208]\tvalid_0's auc: 0.637401\n[209]\tvalid_0's auc: 0.637456\n[210]\tvalid_0's auc: 0.637392\n[211]\tvalid_0's auc: 0.637373\n[212]\tvalid_0's auc: 0.63741\n[213]\tvalid_0's auc: 0.637459\n[214]\tvalid_0's auc: 0.637496\n[215]\tvalid_0's auc: 0.637539\n[216]\tvalid_0's auc: 0.637546\n[217]\tvalid_0's auc: 0.637535\n[218]\tvalid_0's auc: 0.637511\n[219]\tvalid_0's auc: 0.6375\n[220]\tvalid_0's auc: 0.637502\n[221]\tvalid_0's auc: 0.637493\n[222]\tvalid_0's auc: 0.637431\n[223]\tvalid_0's auc: 0.637413\n[224]\tvalid_0's auc: 0.637421\n[225]\tvalid_0's auc: 0.637368\n[226]\tvalid_0's auc: 0.637374\n[227]\tvalid_0's auc: 0.637374\n[228]\tvalid_0's auc: 0.637413\n[229]\tvalid_0's auc: 0.637429\n[230]\tvalid_0's auc: 0.637437\n[231]\tvalid_0's auc: 0.637488\n[232]\tvalid_0's auc: 0.637531\n[233]\tvalid_0's auc: 0.637529\n[234]\tvalid_0's auc: 0.637567\n[235]\tvalid_0's auc: 0.637599\n[236]\tvalid_0's auc: 0.637624\n[237]\tvalid_0's auc: 0.637659\n[238]\tvalid_0's auc: 0.637586\n[239]\tvalid_0's auc: 0.637656\n[240]\tvalid_0's auc: 0.637684\n[241]\tvalid_0's auc: 0.637682\n[242]\tvalid_0's auc: 0.637751\n[243]\tvalid_0's auc: 0.637722\n[244]\tvalid_0's auc: 0.637714\n[245]\tvalid_0's auc: 0.637647\n[246]\tvalid_0's auc: 0.637679\n[247]\tvalid_0's auc: 0.637679\n[248]\tvalid_0's auc: 0.637735\n[249]\tvalid_0's auc: 0.6377\n[250]\tvalid_0's auc: 0.637738\n[251]\tvalid_0's auc: 0.637708\n[252]\tvalid_0's auc: 0.637688\n[253]\tvalid_0's auc: 0.637725\n[254]\tvalid_0's auc: 0.637697\n[255]\tvalid_0's auc: 0.637689\n[256]\tvalid_0's auc: 0.637714\n[257]\tvalid_0's auc: 0.637688\n[258]\tvalid_0's auc: 0.637732\n[259]\tvalid_0's auc: 0.637703\n[260]\tvalid_0's auc: 0.63775\n[261]\tvalid_0's auc: 0.637715\n[262]\tvalid_0's auc: 0.637711\nEarly stopping, best iteration is:\n[242]\tvalid_0's auc: 0.637751\n" ] ], [ [ "## Evaluate Model\n\nEvaluating performance is an essential task in machine learning. In this case, because this is a classification problem, the data scientist elected to use an AUC - ROC Curve. When we need to check or visualize the performance of the multi - class classification problem, we use AUC (Area Under The Curve) ROC (Receiver Operating Characteristics) curve. It is one of the most important evaluation metrics for checking any classification model’s performance.\n\n<img src=\"https://www.researchgate.net/profile/Oxana_Trifonova/publication/276079439/figure/fig2/AS:614187332034565@1523445079168/An-example-of-ROC-curves-with-good-AUC-09-and-satisfactory-AUC-065-parameters.png\"\n alt=\"Markdown Monster icon\"\n style=\"float: left; margin-right: 12px; width: 320px; height: 239px;\" />", "_____no_output_____" ] ], [ [ "predictions = model.predict(valid_data.data)\nfpr, tpr, thresholds = metrics.roc_curve(valid_data.label, predictions)\nmodel_metrics = {\"auc\": (metrics.auc(fpr, tpr))}\nprint(model_metrics)", "{'auc': 0.6377511613946426}\n" ] ], [ [ "## Save Model\n \nIn machine learning, we need to save the trained models in a file and restore them in order to reuse it to compare the model with other models, to test the model on a new data. The saving of data is called Serializaion, while restoring the data is called Deserialization.", "_____no_output_____" ] ], [ [ "model_name = \"lgbm_binary_model.pkl\"\njoblib.dump(value=model, filename=model_name)", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "%matplotlib inline\nimport pandas as pd\nimport numpy as np\nfrom matplotlib import pyplot as plt\nplt.style.use('ggplot')\nimport pickle\n\nfrom sklearn.preprocessing import LabelEncoder\nimport seaborn as sns\ncolor = sns.color_palette()\nsns.set(rc={'figure.figsize':(12,8)})\n\nimport sklearn\nfrom sklearn.preprocessing import MinMaxScaler,StandardScaler,LabelEncoder\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.model_selection import RandomizedSearchCV\nfrom sklearn.metrics import r2_score\nfrom sklearn.model_selection import cross_val_score\nfrom sklearn.linear_model import LinearRegression\nfrom sklearn.linear_model import Ridge\nfrom sklearn.linear_model import Lasso\nfrom sklearn.ensemble import RandomForestRegressor\nfrom sklearn.ensemble import GradientBoostingRegressor\nfrom sklearn import metrics\n\nimport warnings\nwarnings.filterwarnings('ignore')", "_____no_output_____" ], [ "df=pd.read_csv('Cardata-Cleaned.csv',index_col=0)", "_____no_output_____" ], [ "df.info()", "<class 'pandas.core.frame.DataFrame'>\nInt64Index: 1050 entries, 0 to 1049\nData columns (total 17 columns):\n # Column Non-Null Count Dtype \n--- ------ -------------- ----- \n 0 Car Name 1050 non-null object \n 1 Car url 1050 non-null object \n 2 Car Brand 1050 non-null object \n 3 Year 1050 non-null int64 \n 4 Selling Price 1050 non-null float64\n 5 Current Value 1050 non-null float64\n 6 KMs Driven 1050 non-null int64 \n 7 Fuel 1050 non-null object \n 8 Seller Type 1050 non-null object \n 9 max_power 1050 non-null float64\n 10 Transmission 1050 non-null object \n 11 Owner 1050 non-null object \n 12 Mileage 1050 non-null float64\n 13 Engine 1050 non-null int64 \n 14 Drive Type 975 non-null object \n 15 Seats 1050 non-null int64 \n 16 Gear Box 1050 non-null int64 \ndtypes: float64(4), int64(5), object(8)\nmemory usage: 147.7+ KB\n" ] ], [ [ "# Exploratory Data Analysis (EDA)", "_____no_output_____" ], [ "## Univariate Analysis", "_____no_output_____" ] ], [ [ "cat_cols = ['Fuel','Seller Type','Transmission','Owner']\ni=0\nwhile i < 4:\n fig = plt.figure(figsize=[15,6])\n \n plt.subplot(1,2,1)\n sns.countplot(x=cat_cols[i], data=df)\n i += 1\n \n plt.subplot(1,2,2)\n sns.countplot(x=cat_cols[i], data=df)\n i += 1\n \n plt.show()\n", "_____no_output_____" ], [ "num_cols = ['Selling Price','Current Value','KMs Driven','Year','max_power','Mileage','Engine','Gear Box']\ni=0\nwhile i < 8:\n fig = plt.figure(figsize=[15,20])\n plt.subplot(4,2,1)\n sns.boxplot(x=num_cols[i], data=df)\n i += 1\n plt.subplot(4,2,2)\n sns.boxplot(x=num_cols[i], data=df)\n i += 1\n ", "_____no_output_____" ] ], [ [ "## Bivariate Analysis", "_____no_output_____" ] ], [ [ "sns.set(rc={'figure.figsize':(15,15)})\nsns.heatmap(df.corr(),annot=True)", "_____no_output_____" ], [ "print(df['Fuel'].value_counts(),'\\n')\nprint(df['Seller Type'].value_counts(),'\\n')\nprint(df['Transmission'].value_counts(),'\\n')\nprint(df['Owner'].value_counts(),'\\n')", "Petrol 606\nDiesel 442\nElectric 1\nCNG 1\nName: Fuel, dtype: int64 \n\nDealer 722\nIndividual 328\nName: Seller Type, dtype: int64 \n\nManual 905\nAutomatic 145\nName: Transmission, dtype: int64 \n\nFirst Owner 895\nSecond Owner 131\nThird Owner 16\nFourth & Above Owner 8\nName: Owner, dtype: int64 \n\n" ], [ "df.pivot_table(values='Selling Price', index = 'Seller Type', columns= 'Fuel')", "_____no_output_____" ] ], [ [ "# Data Preparation", "_____no_output_____" ], [ "## Creating Dummies for Categorical Features", "_____no_output_____" ] ], [ [ "label_encoder = LabelEncoder()\ndf['Owner']= label_encoder.fit_transform(df['Owner'])", "_____no_output_____" ], [ "final_dataset=df[['Year','Selling Price','Current Value','KMs Driven','Fuel',\n 'Seller Type','max_power','Transmission','Owner','Mileage','Engine','Seats','Gear Box']]", "_____no_output_____" ], [ "final_dataset=pd.get_dummies(final_dataset,drop_first=True)", "_____no_output_____" ], [ "sns.set(rc={'figure.figsize':(15,15)})\nsns.heatmap(final_dataset.corr(),annot=True,cmap=\"RdBu\")", "_____no_output_____" ], [ "final_dataset.corr()['Selling Price'].sort_values(ascending=False)", "_____no_output_____" ], [ "y = final_dataset['Selling Price']\nX = final_dataset.drop('Selling Price',axis=1)", "_____no_output_____" ] ], [ [ "## Feature Importance", "_____no_output_____" ] ], [ [ "from sklearn.ensemble import ExtraTreesRegressor\n\nmodel = ExtraTreesRegressor()\nmodel.fit(X,y)", "_____no_output_____" ], [ "print(model.feature_importances_)", "[1.00727101e-01 4.54874134e-01 1.46965566e-02 1.04696153e-01\n 2.05634981e-03 7.69684617e-03 4.27790659e-02 9.61168352e-02\n 1.14777042e-01 1.01498910e-02 2.35535683e-06 3.42311824e-03\n 7.75940956e-03 4.02451420e-02]\n" ], [ "sns.set(rc={'figure.figsize':(12,8)})\n\nfeat_importances = pd.Series(model.feature_importances_, index=X.columns)\nfeat_importances.nlargest(5).plot(kind='barh')\nplt.show()", "_____no_output_____" ], [ "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=1)\nprint(\"x train: \",X_train.shape)\nprint(\"x test: \",X_test.shape)\nprint(\"y train: \",y_train.shape)\nprint(\"y test: \",y_test.shape)", "x train: (735, 14)\nx test: (315, 14)\ny train: (735,)\ny test: (315,)\n" ], [ "CV = []\nR2_train = []\nR2_test = []\nMAE=[]\nMSE=[]\nRMSE=[]\n\ndef car_pred_model(model):\n \n \n # R2 score of train set\n y_pred_train = model.predict(X_train)\n R2_train_model = r2_score(y_train,y_pred_train)\n R2_train.append(round(R2_train_model,2))\n \n # R2 score of test set\n y_pred_test = model.predict(X_test)\n R2_test_model = r2_score(y_test,y_pred_test)\n R2_test.append(round(R2_test_model,2))\n \n # R2 mean of train set using Cross validation\n cross_val = cross_val_score(model ,X_train ,y_train ,cv=5)\n cv_mean = cross_val.mean()\n CV.append(round(cv_mean,2))\n \n print(\"Train R2-score :\",round(R2_train_model,2))\n print(\"Test R2-score :\",round(R2_test_model,2))\n print(\"Train CV scores :\",cross_val)\n print(\"Train CV mean :\",round(cv_mean,2))\n \n MAE.append(metrics.mean_absolute_error(y_test, y_pred_test))\n MSE.append(metrics.mean_squared_error(y_test, y_pred_test))\n RMSE.append(np.sqrt(metrics.mean_squared_error(y_test, y_pred_test)))\n \n print('MAE:', metrics.mean_absolute_error(y_test, y_pred_test))\n print('MSE:', metrics.mean_squared_error(y_test, y_pred_test))\n print('RMSE:', np.sqrt(metrics.mean_squared_error(y_test, y_pred_test)))\n \n \n fig, ax = plt.subplots(1,2,figsize = (12,6))\n ax[0].set_title('Residual Plot of Train samples')\n sns.distplot((y_test-y_pred_test),hist = True,ax = ax[0])\n ax[0].set_xlabel('y_train - y_pred_train')\n \n # Y_test vs Y_train scatter plot\n ax[1].set_title('y_test vs y_pred_test')\n ax[1].scatter(x = y_test, y = y_pred_test)\n ax[1].set_xlabel('y_test')\n ax[1].set_ylabel('y_pred_test')\n \n plt.show()", "_____no_output_____" ] ], [ [ "## Linear Regression", "_____no_output_____" ] ], [ [ "\nlr = LinearRegression()\nlr.fit(X_train,y_train)\n\ncar_pred_model(lr)", "Train R2-score : 0.83\nTest R2-score : 0.88\nTrain CV scores : [0.87318862 0.23054545 0.81581846 0.86005821 0.79216413]\nTrain CV mean : 0.71\nMAE: 1.3212115374204905\nMSE: 5.731984478802555\nRMSE: 2.3941563187900985\n" ] ], [ [ "## Ridge", "_____no_output_____" ] ], [ [ "\n# Creating Ridge model object\nrg = Ridge()\n# range of alpha \nalpha = np.logspace(-3,3,num=14)\n\n# Creating RandomizedSearchCV to find the best estimator of hyperparameter\nrg_rs = RandomizedSearchCV(estimator = rg, param_distributions = dict(alpha=alpha))\n\nrg_rs.fit(X_train,y_train)\n\ncar_pred_model(rg_rs)", "Train R2-score : 0.83\nTest R2-score : 0.89\nTrain CV scores : [0.87345385 0.19438248 0.81973301 0.87546861 0.79190635]\nTrain CV mean : 0.71\nMAE: 1.259204342611868\nMSE: 5.417232953632579\nRMSE: 2.327494995404411\n" ] ], [ [ "## Lasso", "_____no_output_____" ] ], [ [ "\nls = Lasso()\nalpha = np.logspace(-3,3,num=14) # range for alpha\n\nls_rs = RandomizedSearchCV(estimator = ls, param_distributions = dict(alpha=alpha))\n\nls_rs.fit(X_train,y_train)\ncar_pred_model(ls_rs)", "Train R2-score : 0.83\nTest R2-score : 0.88\nTrain CV scores : [0.87434025 0.22872808 0.82137491 0.87579026 0.79292455]\nTrain CV mean : 0.72\nMAE: 1.28454046198753\nMSE: 5.66135888360262\nRMSE: 2.379361024225332\n" ] ], [ [ "## Random Forest", "_____no_output_____" ] ], [ [ "\nrf = RandomForestRegressor()\n\n# Number of trees in Random forest\nn_estimators=list(range(500,1000,100))\n# Maximum number of levels in a tree\nmax_depth=list(range(4,9,4))\n# Minimum number of samples required to split an internal node\nmin_samples_split=list(range(4,9,2))\n# Minimum number of samples required to be at a leaf node.\nmin_samples_leaf=[1,2,5,7]\n# Number of fearures to be considered at each split\nmax_features=['auto','sqrt']\n\n# Hyperparameters dict\nparam_grid = {\"n_estimators\":n_estimators,\n \"max_depth\":max_depth,\n \"min_samples_split\":min_samples_split,\n \"min_samples_leaf\":min_samples_leaf,\n \"max_features\":max_features}\n\nrf_rs = RandomizedSearchCV(estimator = rf, param_distributions = param_grid,cv = 5, random_state=42, n_jobs = 1)\nrf_rs.fit(X_train,y_train)", "_____no_output_____" ], [ "car_pred_model(rf_rs)", "Train R2-score : 0.96\nTest R2-score : 0.92\nTrain CV scores : [0.90932922 0.55268803 0.77245254 0.92139404 0.90756667]\nTrain CV mean : 0.81\nMAE: 0.796357939896497\nMSE: 3.7897916295929766\nRMSE: 1.9467387163132541\n" ] ], [ [ "## Gradient Boosting", "_____no_output_____" ] ], [ [ "\ngb = GradientBoostingRegressor()\n\n# Rate at which correcting is being made\nlearning_rate = [0.001, 0.01, 0.1, 0.2]\n# Number of trees in Gradient boosting\nn_estimators=list(range(500,1000,100))\n# Maximum number of levels in a tree\nmax_depth=list(range(4,9,4))\n# Minimum number of samples required to split an internal node\nmin_samples_split=list(range(4,9,2))\n# Minimum number of samples required to be at a leaf node.\nmin_samples_leaf=[1,2,5,7]\n# Number of fearures to be considered at each split\nmax_features=['auto','sqrt']\n\n# Hyperparameters dict\nparam_grid = {\"learning_rate\":learning_rate,\n \"n_estimators\":n_estimators,\n \"max_depth\":max_depth,\n \"min_samples_split\":min_samples_split,\n \"min_samples_leaf\":min_samples_leaf,\n \"max_features\":max_features}\n\ngb_rs = RandomizedSearchCV(estimator = gb, param_distributions = param_grid,cv = 5, random_state=42, n_jobs = 1)\n\ngb_rs.fit(X_train,y_train)\n\ncar_pred_model(gb_rs)", "Train R2-score : 0.98\nTest R2-score : 0.92\nTrain CV scores : [0.92462522 0.54253837 0.79010633 0.94300397 0.92394372]\nTrain CV mean : 0.82\nMAE: 0.8893903439168446\nMSE: 3.6312438716414492\nRMSE: 1.905582292015081\n" ], [ "gb_rs.best_params_", "_____no_output_____" ], [ "gb_rs.best_score_", "_____no_output_____" ], [ "Models = [\"Linear Regression\",\"Ridge\",\"Lasso\",\"RandomForest Regressor\",\"GradientBoosting Regressor\"]\n\nscore_comparison=pd.DataFrame({'Model': Models,'R Squared(Train)': R2_train,'R Squared(Test)': R2_test,'CV score mean(Train)': CV,\n 'Mean Absolute Error':MAE,'Mean Squared Error':MSE, 'Root Mean Squared Error':RMSE})\n", "_____no_output_____" ], [ "score_comparison", "_____no_output_____" ], [ "file = open('random_forest_regression_model.pkl', 'wb')\n\n# dump information to that file\npickle.dump(rf_rs, file)", "_____no_output_____" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import os\nimport pickle\n\nfrom neutrinomass.completions import EffectiveOperator, Completion\nfrom neutrinomass.database import ExoticField\nfrom neutrinomass.database import ModelDataFrame, EXOTICS, TERMS, MVDF\nfrom neutrinomass.completions import EFF_OPERATORS\nfrom neutrinomass.completions import DERIV_EFF_OPERATORS", "_____no_output_____" ], [ "DATA_PATH = \"/home/garj/work/neutrinomass/neutrinomass/database\"\nDATA = pickle.load(open(os.path.join(DATA_PATH, \"unfiltered.p\"), \"rb\"))", "_____no_output_____" ], [ "UNF = ModelDataFrame.new(data=DATA, exotics=EXOTICS, terms=TERMS)\n\nSTR_UNF = UNF.drop_duplicates([\"stringent_num\"], keep=\"first\")\nLAGS = len(STR_UNF)\nprint(f\"Number of neutrino-mass mechanisms: {LAGS}\")\n\nDEMO_UNF = UNF.drop_duplicates([\"democratic_num\"], keep=\"first\")\nMODELS = len(DEMO_UNF)\nprint(f\"Number of models: {MODELS}\")\n\nSTR_MVDF = MVDF.drop_duplicates([\"stringent_num\"], keep=\"first\")\nprint(f\"Number of filtered neutrino-mass mechanisms: {len(STR_MVDF)}\")\n\nDEMO_MVDF = MVDF.drop_duplicates([\"democratic_num\"], keep=\"first\")\nprint(f\"Number of filtered neutrino-mass mechanisms: {len(DEMO_MVDF)}\")", "Number of neutrino-mass mechanisms: 430810\nNumber of models: 141989\nNumber of filtered neutrino-mass mechanisms: 11483\nNumber of filtered neutrino-mass mechanisms: 11216\n" ], [ "FIL_DF = MVDF.drop_duplicates(['democratic_num', 'dim'], keep=\"first\")\nUNF_DF = UNF.drop_duplicates(['democratic_num', 'dim'], keep=\"first\")\n\nprint(f\"After filtering, there are {len(FIL_DF[FIL_DF['dim'] == 5])} models derived from dimension-5 operators.\")\nprint(f\"After filtering, there are {len(FIL_DF[FIL_DF['dim'] == 9])} models derived from dimension-9 operators.\")\nprint(f\"After filtering, there are {len(FIL_DF[FIL_DF['dim'] == 11])} models derived from dimension-11 operators.\")\nprint(f\"The total of these is {len(FIL_DF[FIL_DF['dim'] == 5]) + len(FIL_DF[FIL_DF['dim'] == 9]) + len(FIL_DF[FIL_DF['dim'] == 11])}\")\n\nOPS = {**EFF_OPERATORS, **DERIV_EFF_OPERATORS}\nlabels, total, demo, dimensions = [], [], [], []\nfor k in OPS:\n labels.append(k)\n total.append(len(UNF_DF[UNF_DF[\"op\"] == k]))\n demo.append(len(FIL_DF[FIL_DF[\"op\"] == k]))\n dimensions.append(OPS[k].mass_dimension)", "After filtering, there are 3 models derived from dimension-5 operators.\nAfter filtering, there are 244 models derived from dimension-9 operators.\nAfter filtering, there are 10969 models derived from dimension-11 operators.\nThe total of these is 11216\n" ], [ "NHL = STR_UNF.terms[(\"F,00,0,0,0\", \"H\", \"L\")]\nNHSigma = STR_UNF.terms[(\"F,00,2,0,0\", \"H\", \"L\")]\nHHXi1 = STR_UNF.terms[(\"H\", \"H\", \"S,00,2,-1,0\")]\nLLXi1 = STR_UNF.terms[(\"L\", \"L\", \"S,00,2,1,0\")]\n\nN = STR_UNF.exotics[\"F,00,0,0,0\"]\nSigma = STR_UNF.exotics[\"F,00,2,0,0\"]\nXi1 = STR_UNF.exotics[\"S,00,2,1,0\"]", "_____no_output_____" ], [ "N_NHL_lags = len(STR_UNF[STR_UNF[\"stringent_num\"] % NHL == 0])\nN_other_lags = len(STR_UNF[(STR_UNF[\"stringent_num\"] % NHL != 0) & (STR_UNF[\"democratic_num\"] % N == 0)])\n\nSigma_NHSigma_lags = len(STR_UNF[STR_UNF[\"stringent_num\"] % NHSigma == 0])\nSigma_other_lags = len(STR_UNF[(STR_UNF[\"stringent_num\"] % NHSigma != 0) & (STR_UNF[\"democratic_num\"] % Sigma == 0)])\n\nXi1_HHXi1_lags = len(STR_UNF[STR_UNF[\"stringent_num\"] % HHXi1 == 0])\nXi1_LLXi1_lags = len(STR_UNF[STR_UNF[\"stringent_num\"] % LLXi1 == 0])\nXi1_both_lags = len(STR_UNF[(STR_UNF[\"stringent_num\"] % HHXi1 == 0) & (STR_UNF[\"stringent_num\"] % LLXi1 == 0)])\nXi1_other_lags = len(STR_UNF[(STR_UNF[\"stringent_num\"] % HHXi1 != 0) & (STR_UNF[\"stringent_num\"] % LLXi1 != 0) & (STR_UNF[\"democratic_num\"] % Xi1 == 0)])", "_____no_output_____" ], [ "N_models = len(DEMO_UNF[DEMO_UNF[\"democratic_num\"] % N == 0])\nSigma_models = len(DEMO_UNF[DEMO_UNF[\"democratic_num\"] % Sigma == 0])\nXi1_models = len(DEMO_UNF[DEMO_UNF[\"democratic_num\"] % Xi1 == 0])", "_____no_output_____" ], [ "# latex table\nprint(r\"\"\"\n \\begin{tabular}{ccll}\n \\toprule\n Field & Interactions & Lagrangians & Collected models \\\\\n \\midrule\n \\multirow{2}{*}{$N \\sim (\\mathbf{1}, \\mathbf{1}, 0)_{F}$} & $L H N$ & %s (%s) & \\multirow{2}{*}{%s (%s)} \\\\\n & Other & %s (%s) & \\\\\n \\midrule\n \\multirow{2}{*}{$\\Sigma \\sim (\\mathbf{1}, \\mathbf{3}, 0)_{F}$} & $L H \\Sigma$ & %s (%s) & \\multirow{2}{*}{%s (%s)} \\\\\n & Other & %s (%s) & \\\\\n \\midrule\n \\multirow{4}{*}{$\\Xi_{1} \\sim (\\mathbf{1}, \\mathbf{3}, 1)_{S}$} & $L L \\Xi_{1}$ & %s (%s) & \\multirow{4}{*}{%s (%s)} \\\\\n & $H H \\Xi_{1}^{\\dagger}$ & %s (%s) & \\\\\n & Both & %s (%s) & \\\\\n & Other & %s (%s) & \\\\\n \\bottomrule\n \\end{tabular}\n\"\"\" % (\n f\"{N_NHL_lags:,}\", f\"{100 * N_NHL_lags / LAGS:.1f}\\%\",\n f\"{N_models:,}\", f\"{100 * N_models / MODELS:.1f}\\%\",\n f\"{N_other_lags:,}\", f\"{100 * N_other_lags / LAGS:.1f}\\%\",\n\n \n f\"{Sigma_NHSigma_lags:,}\", f\"{100 * Sigma_NHSigma_lags / LAGS:.1f}\\%\",\n f\"{Sigma_models:,}\", f\"{100 * Sigma_models / MODELS:.1f}\\%\",\n f\"{Sigma_other_lags:,}\", f\"{100 * Sigma_other_lags / LAGS:.1f}\\%\",\n\n \n f\"{Xi1_LLXi1_lags:,}\", f\"{100 * Xi1_LLXi1_lags / LAGS:.1f}\\%\",\n f\"{Xi1_models:,}\", f\"{100 * Xi1_models / MODELS:.1f}\\%\",\n f\"{Xi1_HHXi1_lags:,}\", f\"{100 * Xi1_HHXi1_lags / LAGS:.1f}\\%\",\n f\"{Xi1_both_lags:,}\", f\"{100 * Xi1_both_lags / LAGS:.1f}\\%\",\n f\"{Xi1_other_lags:,}\", f\"{100 * Xi1_other_lags / LAGS:.1f}\\%\",\n)\n)", "\n \\begin{tabular}{ccll}\n \\toprule\n Field & Interactions & Lagrangians & Collected models \\\\\n \\midrule\n \\multirow{2}{*}{$N \\sim (\\mathbf{1}, \\mathbf{1}, 0)_{F}$} & $L H N$ & 51,245 (11.9\\%) & \\multirow{2}{*}{17,139 (12.1\\%)} \\\\\n & Other & 12,433 (2.9\\%) & \\\\\n \\midrule\n \\multirow{2}{*}{$\\Sigma \\sim (\\mathbf{1}, \\mathbf{3}, 0)_{F}$} & $L H \\Sigma$ & 87,535 (20.3\\%) & \\multirow{2}{*}{31,629 (22.3\\%)} \\\\\n & Other & 28,157 (6.5\\%) & \\\\\n \\midrule\n \\multirow{4}{*}{$\\Xi_{1} \\sim (\\mathbf{1}, \\mathbf{3}, 1)_{S}$} & $L L \\Xi_{1}$ & 59,791 (13.9\\%) & \\multirow{4}{*}{51,576 (36.3\\%)} \\\\\n & $H H \\Xi_{1}^{\\dagger}$ & 95,410 (22.1\\%) & \\\\\n & Both & 10,323 (2.4\\%) & \\\\\n & Other & 30,761 (7.1\\%) & \\\\\n \\bottomrule\n \\end{tabular}\n\n" ], [ "import matplotlib.pyplot as plt\nimport seaborn as sns\nimport pandas as pd\n\nSMALL_SIZE = 15\nMEDIUM_SIZE = 20\nBIGGER_SIZE = 20\n\nplt.rc('font', size=SMALL_SIZE) # controls default text sizes\nplt.rc('axes', titlesize=SMALL_SIZE) # fontsize of the axes title\nplt.rc('axes', labelsize=MEDIUM_SIZE) # fontsize of the x and y labels\nplt.rc('xtick', labelsize=SMALL_SIZE) # fontsize of the tick labels\nplt.rc('ytick', labelsize=SMALL_SIZE) # fontsize of the tick labels\nplt.rc('legend', fontsize=SMALL_SIZE) # legend fontsize\nplt.rc('figure', titlesize=BIGGER_SIZE) # fontsize of the figure title\n\nplt.tight_layout()\n\nplt.rcParams.update({\n \"text.usetex\": True,\n \"font.family\": \"serif\",\n \"font.serif\": [\"Computer Modern Roman\"]}\n)\n\nsns.set_palette(\"muted\")", "_____no_output_____" ], [ "latex_labels = []\nfor l in labels:\n if \"pp\" in l:\n new_l = l.replace(\"pp\", \"^{\\prime\\prime}\")\n elif \"p\" in l:\n new_l = l.replace(\"p\", \"^\\prime\")\n else:\n new_l = l\n latex_labels.append(\"$\" + new_l + \"$\")\n\nfilter_bar_df = pd.DataFrame(data={\n \"Operator\": latex_labels, \n \"Unfiltered\": total, \n \"Democratic\": demo,\n \"Dimension\": dimensions\n})\n\ndemo_5 = sum(filter_bar_df[filter_bar_df[\"Dimension\"] == 5][\"Democratic\"])\ndemo_7 = sum(filter_bar_df[filter_bar_df[\"Dimension\"] == 7][\"Democratic\"])\ndemo_9 = sum(filter_bar_df[filter_bar_df[\"Dimension\"] == 9][\"Democratic\"])\ndemo_11 = sum(filter_bar_df[filter_bar_df[\"Dimension\"] == 11][\"Democratic\"])\n\n\nunf_5 = sum(filter_bar_df[filter_bar_df[\"Dimension\"] == 5][\"Unfiltered\"])\nunf_7 = sum(filter_bar_df[filter_bar_df[\"Dimension\"] == 7][\"Unfiltered\"])\nunf_9 = sum(filter_bar_df[filter_bar_df[\"Dimension\"] == 9][\"Unfiltered\"])\nunf_11 = sum(filter_bar_df[filter_bar_df[\"Dimension\"] == 11][\"Unfiltered\"])\n\nbarplot_df = pd.DataFrame(\n {'Dimension': [5, 7, 9, 11], \n 'Democratic': [demo_5, demo_7, demo_9, demo_11], \n 'Unfiltered': [unf_5-demo_5, unf_7-demo_7, unf_9-demo_9, unf_11-demo_11]}\n)", "_____no_output_____" ], [ "ax = barplot_df.plot.bar(x=\"Dimension\", stacked=True, rot=0)\nax.set_yscale(\"log\")\nax.set_ylabel(\"Number of models\")\nplt.tight_layout()\nplt.savefig(\"/home/garj/filter_barchart_dimension.pdf\")\nplt.savefig(\"/home/garj/filter_barchart_dimension.png\")", "_____no_output_____" ], [ "ops_filter_bar_df = filter_bar_df[filter_bar_df[\"Dimension\"] < 11]\n\nf, ax = plt.subplots(figsize=(7, 10))\n\nsns.barplot(x=\"Unfiltered\", y=\"Operator\", data=ops_filter_bar_df, label=\"Unfiltered\", color=sns.color_palette()[1])\nsns.barplot(x=\"Democratic\", y=\"Operator\", data=ops_filter_bar_df, label=\"Democratic\", color=sns.color_palette()[0])\nax.set_xscale(\"log\")\nax.legend(ncol=2, loc=\"upper right\", frameon=True)\nax.set(xlim=(0, 10000), ylabel=\"Operator\", xlabel=\"Number of models\")\n\nax.text(x=2000, y=7, s=\"$d < 11$\", fontsize=20)\n \nfor tick in ax.yaxis.get_major_ticks()[1::2]:\n tick.set_pad(40)\n\nplt.tight_layout()\nplt.savefig(\"/home/garj/filter_barchart_operators579.pdf\")\nplt.savefig(\"/home/garj/filter_barchart_operators579.png\")", "/usr/local/easybuild-2019/easybuild/software/mpi/gcc/8.3.0/openmpi/3.1.4/jupyter/1.0.0-python-3.7.4/lib/python3.7/site-packages/ipykernel_launcher.py:9: UserWarning: Attempted to set non-positive left xlim on a log-scaled axis.\nInvalid limit will be ignored.\n if __name__ == '__main__':\n" ], [ "import seaborn as sns\nimport matplotlib.pyplot as plt\n\nops_filter_bar_df = filter_bar_df[filter_bar_df[\"Dimension\"] == 11]\n\nf, ax = plt.subplots(figsize=(7, 15))\n\nsns.barplot(x=\"Unfiltered\", y=\"Operator\", data=ops_filter_bar_df, label=\"Unfiltered\", color=sns.color_palette()[1])\nsns.barplot(x=\"Democratic\", y=\"Operator\", data=ops_filter_bar_df, label=\"Democratic\", color=sns.color_palette()[0])\nax.set_xscale(\"log\")\nax.legend(ncol=2, loc=\"upper right\", frameon=True)\nax.set(xlim=(0, 100000), ylabel=\"Operator\", xlabel=\"Number of models\")\n\nax.text(x=12000, y=9, s=\"$d = 11$\", fontsize=22)\n \nfor tick in ax.yaxis.get_major_ticks()[1::2]:\n tick.set_pad(40)\n\nplt.tight_layout()\nplt.savefig(\"/home/garj/filter_barchart_operators11.pdf\")\nplt.savefig(\"/home/garj/filter_barchart_operators11.png\")", "/usr/local/easybuild-2019/easybuild/software/mpi/gcc/8.3.0/openmpi/3.1.4/jupyter/1.0.0-python-3.7.4/lib/python3.7/site-packages/ipykernel_launcher.py:12: UserWarning: Attempted to set non-positive left xlim on a log-scaled axis.\nInvalid limit will be ignored.\n if sys.path[0] == '':\n" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import numpy as np\nimport pandas as pd \nimport matplotlib.pyplot as plt", "_____no_output_____" ], [ "df=pd.read_csv('data_1000.csv')", "_____no_output_____" ], [ "data=df[['correct_answ','bleu_score','levenstein_sim','cosine_sim','jaccard_sim']]", "_____no_output_____" ], [ "data.head(10)", "_____no_output_____" ], [ "data.describe()", "_____no_output_____" ], [ "data.boxplot(by='correct_answ', column=['bleu_score', 'levenstein_sim', 'cosine_sim', 'jaccard_sim'], \n grid=True, figsize=(15,15))", "_____no_output_____" ], [ "X=data[['bleu_score','levenstein_sim','cosine_sim','jaccard_sim']]\ny=data['correct_answ']", "_____no_output_____" ], [ "X.hist(bins=50,figsize=(20,15))", "_____no_output_____" ], [ "from pandas.plotting import scatter_matrix\nscatter_matrix(X, figsize=(14, 10))", "_____no_output_____" ], [ "corr_matrix = data.corr()\ncorr_matrix[\"correct_answ\"].sort_values(ascending=False)", "_____no_output_____" ], [ "data['cos_lev']=data['cosine_sim']+data['levenstein_sim']\ndata['cos_bleu']=data['cosine_sim']+data['bleu_score']\ndata['cos_jac']=data['cosine_sim']+data['jaccard_sim']\ndata['lev_bleu']=data['levenstein_sim']+data['bleu_score']\ndata['lev_jac']=data['levenstein_sim']+data['jaccard_sim']\ndata['bleu_jac']=data['bleu_score']+data['jaccard_sim']", "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/ipykernel_launcher.py:1: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n \"\"\"Entry point for launching an IPython kernel.\n/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/ipykernel_launcher.py:2: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n \n/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/ipykernel_launcher.py:3: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n This is separate from the ipykernel package so we can avoid doing imports until\n/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/ipykernel_launcher.py:4: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n after removing the cwd from sys.path.\n/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/ipykernel_launcher.py:5: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n \"\"\"\n/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/ipykernel_launcher.py:6: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n \n" ], [ "corr_matrix = data.corr()\ncorr_matrix[\"correct_answ\"].sort_values(ascending=False)\ncorr_matrix", "_____no_output_____" ], [ "X=data[['bleu_score','levenstein_sim','cosine_sim','jaccard_sim','cos_lev','cos_bleu','cos_jac','lev_bleu','lev_jac','bleu_jac']]\ny=data[['correct_answ']]", "_____no_output_____" ], [ "from sklearn.model_selection import StratifiedShuffleSplit\n\nsplit = StratifiedShuffleSplit(n_splits=1, test_size=0.2, random_state=42)\n\nfor train_index, test_index in split.split(data, data[\"correct_answ\"]):\n strat_train_set = data.loc[train_index]\n strat_test_set = data.loc[test_index]", "_____no_output_____" ], [ "data[\"correct_answ\"].value_counts() / len(data)", "_____no_output_____" ], [ "strat_train_set[\"correct_answ\"].value_counts() / len(strat_train_set)", "_____no_output_____" ], [ "strat_test_set[\"correct_answ\"].value_counts() / len(strat_test_set)", "_____no_output_____" ], [ "training = strat_train_set.copy()\ntesting=strat_test_set.copy()", "_____no_output_____" ], [ "X_training=training[['bleu_score','levenstein_sim','cosine_sim','jaccard_sim','cos_lev','cos_bleu','cos_jac','lev_bleu','lev_jac','bleu_jac']]\ny_training=training[['correct_answ']]\nX_testing=testing[['bleu_score','levenstein_sim','cosine_sim','jaccard_sim','cos_lev','cos_bleu','cos_jac','lev_bleu','lev_jac','bleu_jac']]\ny_testing=testing[['correct_answ']]\n", "_____no_output_____" ], [ "from sklearn.neighbors import KNeighborsClassifier\nclf = KNeighborsClassifier(n_neighbors=2).fit(X_training,y_training)\ny_training_prediction = clf.predict(X_training)\ny_testing_prediction = clf.predict(X_testing)", "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/ipykernel_launcher.py:2: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n \n" ], [ "#K_fold confusion_matrix for training sets & Precision score & Recall score\n\nfrom sklearn.model_selection import StratifiedKFold\nfrom sklearn.metrics import precision_score, recall_score\nfrom sklearn.base import clone\nfrom sklearn.metrics import confusion_matrix\nfrom sklearn.metrics import f1_score\nskfolds = StratifiedKFold(n_splits=3, random_state=42)\nfor train_index, test_index in skfolds.split(X_training, y_training):\n clone_clf = clone(clf)\n X_train_folds = X_training.iloc[train_index]\n y_train_folds = y_training.iloc[train_index]\n X_test_fold = X_training.iloc[test_index]\n y_test_fold = y_training.iloc[test_index]\n clone_clf.fit(X_train_folds, y_train_folds)\n y_pred = clone_clf.predict(X_test_fold)\n recall_score(y_test_fold,y_pred)\n print(confusion_matrix(y_test_fold,y_pred))\n print('precision_score: ',precision_score(y_test_fold,y_pred))\n print('recall_score: ',recall_score(y_test_fold,y_pred))\n print('f1_socre: ',f1_score(y_test_fold, y_pred))\n print('')\n print('\\n')\n \n\"\"\" n_correct = sum(y_pred == y_test_fold)\n print(n_correct / len(y_pred)) \"\"\"\n", "[[ 54 3]\n [ 10 205]]\nprecision_score: 0.9855769230769231\nrecall_score: 0.9534883720930233\nf1_socre: 0.9692671394799054\n\n\n\n[[ 54 2]\n [ 18 197]]\nprecision_score: 0.9899497487437185\nrecall_score: 0.9162790697674419\nf1_socre: 0.9516908212560387\n\n\n\n[[ 55 2]\n [ 20 194]]\nprecision_score: 0.9897959183673469\nrecall_score: 0.9065420560747663\nf1_socre: 0.9463414634146342\n\n\n\n" ], [ "# cross_validation score for training sets\nfrom sklearn.model_selection import cross_val_score\ncross_val_score(clf, X_training, y_training, cv=3, scoring=\"accuracy\")", "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/sklearn/model_selection/_validation.py:515: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n estimator.fit(X_train, y_train, **fit_params)\n/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/sklearn/model_selection/_validation.py:515: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n estimator.fit(X_train, y_train, **fit_params)\n/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/sklearn/model_selection/_validation.py:515: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n estimator.fit(X_train, y_train, **fit_params)\n" ], [ "# testing data validation\nfrom sklearn.metrics import confusion_matrix\nprint('confusion_matrix: \\n',confusion_matrix(y_testing,y_testing_prediction))\nprint('precision_score: ',precision_score(y_testing,y_testing_prediction))\nprint('recall_score: ',recall_score(y_testing,y_testing_prediction))\nprint('f1_socre: ',f1_score(y_testing,y_testing_prediction))\nfrom sklearn.model_selection import cross_val_score\nprint('accuracy: ',cross_val_score(clf, X_testing, y_testing, cv=3, scoring=\"accuracy\"))", "confusion_matrix: \n [[ 40 2]\n [ 12 150]]\nprecision_score: 0.9868421052631579\nrecall_score: 0.9259259259259259\nf1_socre: 0.9554140127388535\naccuracy: [0.82352941 0.79411765 0.91176471]\n" ], [ "import matplotlib.pyplot ", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]