# -*- coding: utf-8 -*-
"""Lab2222.ipynb

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/1OUGOeTdmMbccW_st3Ao8nHDR5wm_VUNg


from google.colab import drive

drive.mount("/content/ML_Course")

cd /content/ML_Course/MyDrive/ML_Course
"""
pip3 install -U scikit-learn scipy matplotlib
import pandas as pd
housing = pd.read_csv("housing.csv")
housing.head(n = 5)

housing.columns

housing.describe()

housing.info()

# Commented out IPython magic to ensure Python compatibility.
# %matplotlib inline
import matplotlib.pyplot as plt
housing.hist(bins=50, figsize=(20,15))
plt.show()

# to make this notebook's output identical at every run
import numpy as np
np.random.seed(10)

# For illustration only. Sklearn has train_test_split()
def split_train_test(data, test_ratio):
    shuffled_indices = np.random.permutation(len(data))
    test_set_size = int(len(data) * test_ratio)
    test_indices = shuffled_indices[:test_set_size]
    train_indices = shuffled_indices[test_set_size:]
    return data.iloc[train_indices], data.iloc[test_indices]

# run the function to get the train & test set
train_set, test_set = split_train_test(housing, 0.2)

train_set.info()

test_set.info()

from sklearn.model_selection import train_test_split
train_set, test_set = train_test_split(housing, test_size=0.2, random_state=10)

train_set.info()

test_set.info()

test_set.to_csv('blind_test.csv', index = False)

train_set.plot(kind="scatter", x="longitude", y="latitude", alpha=0.4, 
    s=train_set["population"]/100, label="population", figsize=(10,7),
    c="median_house_value", cmap=plt.get_cmap("jet"), colorbar=True,
    sharex=False)
plt.legend()
plt.show()

train_set.info()

train_set[train_set.isna().any(axis=1)]

train_set_clean = train_set.dropna(subset=["total_bedrooms"])
train_set_clean

train_set_clean.info()

train_labels = train_set_clean["median_house_value"].copy() # get labels for output label Y
train_features = train_set_clean.drop("median_house_value", axis=1) # drop labels to get features X for training set
train_features.info()

train_features.head()

train_features.columns

train_features.info()

train_features.describe()

train_labels

train_features.hist(bins=50, figsize=(12,9))

train_features.describe()

from sklearn.preprocessing import MinMaxScaler
scaler = MinMaxScaler() ## define the transformer
scaler.fit(train_features) ## call .fit() method to calculate the min and max value for each column in dataset

print("Min of each column: ",scaler.data_min_)
print("Max of each column: ",scaler.data_max_)

train_features.describe()

train_features_normalized = scaler.transform(train_features)
train_features_normalized

pd.DataFrame(train_features_normalized).hist(bins=50, figsize=(12,9))
plt.show()

## 1. split data to get train and test set
from sklearn.model_selection import train_test_split
train_set, test_set = train_test_split(housing, test_size=0.2, random_state=10)

## 2. clean the missing values
train_set_clean = train_set.dropna(subset=["total_bedrooms"])
train_set_clean

## 2. derive training features and training labels 
train_labels = train_set_clean["median_house_value"].copy() # get labels for output label Y
train_features = train_set_clean.drop("median_house_value", axis=1) # drop labels to get features X for training set


## 4. scale the numeric features in training set
from sklearn.preprocessing import MinMaxScaler
scaler = MinMaxScaler() ## define the transformer
scaler.fit(train_features) ## call .fit() method to calculate the min and max value for each column in dataset

train_features_normalized = scaler.transform(train_features)
train_features_normalized

from sklearn.linear_model import LinearRegression ## import the LinearRegression Function
lin_reg = LinearRegression() ## Initialize the class
lin_reg.fit(train_features_normalized, train_labels) # feed the training data X, and label Y for supervised learning
# feed the training data X, and label Y for supervised learning

training_predictions = lin_reg.predict(train_features_normalized)
training_predictions.shape

train_labels

## plot scatter plot 
import matplotlib.pyplot as plt
plt.scatter(training_predictions, train_labels )
plt.xlabel('training_predictions', fontsize=15,color="red")
plt.ylabel('train_label', fontsize=15,color="green")
plt.title('Scatter plot for training_predictions and train_label', fontsize=15)
plt.xlim(0,np.max(training_predictions)) # remove the predictions that have negative prices
plt.show()

import numpy as np
np.corrcoef(training_predictions, train_labels)

import pandas as pd 
prediction_summary = pd.DataFrame({'predicted_label':training_predictions, 'actual_label':train_labels})
prediction_summary

prediction_summary['error'] = prediction_summary['actual_label'] - prediction_summary['predicted_label']
prediction_summary

from sklearn.metrics import mean_squared_error
lin_mse = mean_squared_error(train_labels, training_predictions)
lin_rmse = np.sqrt(lin_mse)
lin_rmse

## Step 1: training the data using decision tree algorithm
from sklearn.tree import DecisionTreeRegressor ## import the DecisionTree Function
tree_reg = DecisionTreeRegressor(random_state=10) ## Initialize the class
tree_reg.fit(train_features_normalized, train_labels) # feed the training data X, and label Y for supervised learning

### Step 2: make a prediction using tree model
training_predictions_trees = tree_reg.predict(train_features_normalized)
training_predictions_trees

## Step 3: visualize the scatter plot between predictions and actual labels
import matplotlib.pyplot as plt
plt.scatter(training_predictions_trees, train_labels )
plt.xlabel('training_predictions_trees', fontsize=15,color="red")
plt.ylabel('train_label', fontsize=15,color="green")
plt.title('Scatter plot for training_predictions_trees and train_label', fontsize=15)
plt.xlim(0,np.max(training_predictions_trees)) # remove the predictions that have negative prices
plt.show()

from sklearn.metrics import mean_squared_error
tree_mse = mean_squared_error(train_labels, training_predictions_trees)
tree_rmse = np.sqrt(tree_mse)
tree_rmse

## 1. clean the missing values in test set
test_set_clean = test_set.dropna(subset=["total_bedrooms"])
test_set_clean

## 2. derive test features and test labels. In this case, test labels are only used for evaluation
test_labels = test_set_clean["median_house_value"].copy() # get labels for output label Y
test_features = test_set_clean.drop("median_house_value", axis=1) # drop labels to get features X for training set


## 4. scale the numeric features in test set. 
## important note: do not apply fit function on the test set, using same scalar from training set
test_features_normalized = scaler.transform(test_features)
test_features_normalized

### Step 5: make a prediction using tree model
test_predictions_trees = tree_reg.predict(test_features_normalized)
test_predictions_trees

from sklearn.metrics import mean_squared_error
test_tree_mse = mean_squared_error(test_labels, test_predictions_trees)
test_tree_rmse = np.sqrt(test_tree_mse)
test_tree_rmse

# Step 1: install Gradio
#!pip install --quiet gradio

# Step 2: import library
#import gradio as gr
#print(gr.__version__)

# Step 3.1: Define a simple "Hello World" function
# requirement: input is text, output is text
def greet(name):
      return "Hello " + name + "!!"

# Step 3.2: Define the input component (text style) and output component (text style) to create a simple GUI
import gradio as gr
input_module = gr.inputs.Textbox(label = "Input Text")
output_module = gr.outputs.Textbox(label = "Output Text")

# Step 3.3: Put all three component together into the gradio's interface function 
gr.Interface(fn=greet, inputs=input_module, outputs=output_module).launch()

# Step 5.1: Define a simple "image-to-text" function
# requirement: input is text, output is text

def caption(image):
    return "Image is processed!!"

# Step 5.2: Define the input component (image style) and output component (text style) to create a simple GUI
import gradio as gr
input_module = gr.inputs.Image(label = "Input Image")

output_module = gr.outputs.Textbox(label = "Output Text")

# Step 5.3: Put all three component together into the gradio's interface function 
gr.Interface(fn=caption, inputs=input_module, outputs=output_module).launch()

# Step 6.1: Define different input components
import gradio as gr

# a. define text data type
input_module1 = gr.inputs.Textbox(label = "Input Text")

# b. define image data type
input_module2 = gr.inputs.Image(label = "Input Image")

# c. define Number data type
input_module3 = gr.inputs.Number(label = "Input Number")

# d. define Slider data type
input_module4 = gr.inputs.Slider(1, 100, step=5, label = "Input Slider")

# e. define Checkbox data type
input_module5 = gr.inputs.Checkbox(label = "Does it work?")

# f. define Radio data type
input_module6 = gr.inputs.Radio(choices=["park", "zoo", "road"], label = "Input Radio")

# g. define Dropdown data type
input_module7 = gr.inputs.Dropdown(choices=["park", "zoo", "road"], label = "Input Dropdown")

# Step 6.2: Define different output components
# a. define text data type
output_module1 = gr.outputs.Textbox(label = "Output Text")

# b. define image data type
output_module2 = gr.outputs.Image(label = "Output Image")

# you can define more output components

# Step 6.3: Define a new function that accommodates the input modules.
def multi_inputs(input1, input2, input3, input4, input5, input6, input7 ):
    import numpy as np
    ## processing inputs

    ## return outputs
    output1 = "Processing inputs and return outputs" # text output example
    output2 = np.random.rand(6,6) # image-like array output example
    return output1,output2

# Step 6.4: Put all three component together into the gradio's interface function 
gr.Interface(fn=multi_inputs, 
             inputs=[input_module1, input_module2, input_module3,
                     input_module4, input_module5, input_module6,
                     input_module7], 
             outputs=[output_module1, output_module2]
            ).launch()

# Step 6.1: Define different input components
import gradio as gr

# a. define text data type
input_module1 = gr.inputs.Slider(-124.35,-114.35, step =0.5,label = "Longitude")

# b. define image data type
input_module2 = gr.inputs.Slider(32,41, step =0.5,label = "Latitude")

# c. define Number data type
input_module3 = gr.inputs.Slider(1,52, step = 1,label = "Housing_median_age(Year)")

# d. define Slider data type
input_module4 = gr.inputs.Slider(1, 40000, step=1, label = "Total_rooms")

# e. define Checkbox data type
input_module5 = gr.inputs.Slider(1, 6441,label = "Total_bedrooms")

# f. define Radio data type
input_module6 = gr.inputs.Slider(1,6441,step = 1,label = "Population")

# g. define Dropdown data type
input_module7 = gr.inputs.Slider(1,6081,step = 1,label = "Households")

input_module8 = gr.inputs.Slider(0,15,step = 1,label = "Median_income")

# Step 6.2: Define different output components
# a. define text data type
output_module1 = gr.outputs.Textbox(label = "Predicted Housing Prices")

# b. define image data type
output_module2 = gr.outputs.Image(label = "Output Image")

# you can define more output components

train_set.columns

#save machinel earning model to local drive
import pickle
#save 
with open('tree_reg.pkl','wb') as f:
  pickle.dump(tree_reg,f)

ls

# Step 6.3: Define a new function that accommodates the input modules.
def machine_learning_model(input1, input2, input3, input4, input5, input6, input7, input8):
    print('Start ML process')
    import numpy as np
    import pandas as pd
    print(input1, input2, input3, input4, input5, input6, input7, input8)
    #1. process the user submission
    new_feature = np.array([[input1, input2, input3, input4, input5, input6, input7, input8]])
    print(new_feature)
    
    test_set = pd.DataFrame(new_feature, columns = ['longitude', 'latitude', 'housing_median_age', 'total_rooms',
       'total_bedrooms', 'population', 'households', 'median_income'])
    
    ## 1. clean the missing values in test set
    test_set_clean = test_set.dropna(subset=["total_bedrooms"])
    test_set_clean

    ## 2. derive test features and test labels. In this case, test labels are only used for evaluation
    #test_labels = test_set_clean["median_house_value"].copy() # get labels for output label Y
    #test_features = test_set_clean.drop("median_house_value", axis=1) # drop labels to get features X for training set

    test_features_normalized = scaler.transform(test_set_clean)
    print("test_features_normalized: ", test_features_normalized)
    
    with open('tree_reg.pkl','rb') as f:
      tree_reg = pickle.load(f)
    print("Start processing")

    output1 = 'This is the output'
    output2 = np.random.rand(28,28)

    #2. follow the data preprocessing steps as we have done in the test data
    #2.2 Check missing values in total_bedrroms
    # 2.2 feature normalization

    #3. load pre trained machine learning

    
    #4 apply loaded modeld
    test_predictions_trees = tree_reg.predict(test_features_normalized)
    print("Predicition is :",test_predictions_trees)
    
    import matplotlib.pyplot as plt

    train_set.plot(kind="scatter", x="longitude", y="latitude", alpha=0.4, 
    s=train_set["population"]/100, label="population", figsize=(10,7),
    c="median_house_value", cmap=plt.get_cmap("jet"), colorbar=True,
    sharex=False)
    plt.legend()
    
    #plt.show()
    plt.xlim(-124.35,-114.35)
    plt.ylim(32,41)
    plt.plot([input1],[input2],marker = "X",markersize = 20, markeredgecolor="yellow", markerfacecolor="black")
    plt.savefig('test.png')
    #5 send back the prediciton
    return test_predictions_trees,'test.png'

gr.Interface(fn=machine_learning_model, 
             inputs=[input_module1, input_module2, input_module3,
                     input_module4, input_module5, input_module6,
                     input_module7, input_module8], 
             outputs=[output_module1, output_module2]
            ).launch(debug = True)