import os import numpy as np import matplotlib.pyplot as plt import gradio as gr import pandas as pd import tarfile import urllib.request DOWNLOAD_ROOT = "https://raw.githubusercontent.com/ageron/handson-ml2/master/" HOUSING_PATH = os.path.join("datasets", "housing") HOUSING_URL = DOWNLOAD_ROOT + "datasets/housing/housing.tgz" def fetch_housing_data(housing_url=HOUSING_URL, housing_path=HOUSING_PATH): if not os.path.isdir(housing_path): os.makedirs(housing_path) tgz_path = os.path.join(housing_path, "housing.tgz") urllib.request.urlretrieve(housing_url, tgz_path) housing_tgz = tarfile.open(tgz_path) housing_tgz.extractall(path=housing_path) housing_tgz.close() def load_housing_data(housing_path=HOUSING_PATH): csv_path = os.path.join(housing_path, "housing.csv") return pd.read_csv(csv_path) #1. Download the data fetch_housing_data() housing_pd = load_housing_data() housing_pd.head() ## tentatively drop categorical feature housing = housing_pd.drop('ocean_proximity', axis=1) housing #2. Prepare the Data for Machine Learning Algorithms ## 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 #3. Training ML model on the Training Set 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 ### visualize the data def save_fig(fig_id, tight_layout=True, fig_extension="png", resolution=300): path = os.path.join(IMAGES_PATH, fig_id + "." + fig_extension) print("Saving figure", fig_id, ' to ',path) if tight_layout: plt.tight_layout() plt.savefig(path, format=fig_extension, dpi=resolution) PROJECT_ROOT_DIR='./' IMAGES_PATH = os.path.join(PROJECT_ROOT_DIR, "images") os.makedirs(IMAGES_PATH, exist_ok=True) images_path = os.path.join(PROJECT_ROOT_DIR, "images", "end_to_end_project") os.makedirs(images_path, exist_ok=True) DOWNLOAD_ROOT = "https://raw.githubusercontent.com/ageron/handson-ml2/master/" filename = "california.png" print("Downloading", filename) url = DOWNLOAD_ROOT + "images/end_to_end_project/" + filename urllib.request.urlretrieve(url, os.path.join(images_path, filename)) ### written by Jie def draw_map_customize(longitude,latitude, fig_id='test',fig_extension='png' ): import matplotlib.image as mpimg california_img=mpimg.imread(os.path.join(images_path, filename)) ax = housing.plot(kind="scatter", x="longitude", y="latitude", figsize=(10,7), s=housing['population']/100, label="Population", c="median_house_value", cmap=plt.get_cmap("jet"), colorbar=False, alpha=0.4) plt.imshow(california_img, extent=[-124.55, -113.80, 32.45, 42.05], alpha=0.5, cmap=plt.get_cmap("jet")) plt.ylabel("Latitude", fontsize=18) plt.xlabel("Longitude", fontsize=18) plt.xticks(fontsize=18, rotation=0) plt.yticks(fontsize=18, rotation=0) plt.plot(longitude,latitude, "ro", alpha=0.7, marker=r'$\clubsuit$', markersize=30) plt.annotate("Your location is here", xy=(longitude,latitude), xytext=(longitude+1,latitude+1), fontsize=20, arrowprops=dict(arrowstyle="->")) prices = housing["median_house_value"] tick_values = np.linspace(prices.min(), prices.max(), 11) cbar = plt.colorbar(ticks=tick_values/prices.max()) cbar.ax.set_yticklabels(["$%dk"%(round(v/1000)) for v in tick_values], fontsize=14) cbar.set_label('Median House Value', fontsize=16) plt.legend(fontsize=16) save_fig(fig_id) #plt.show() path = os.path.join(IMAGES_PATH, fig_id + "." + fig_extension) return path def get_sample_data(num_data): sample_data = [] for i in range(num_data): samp = housing.sample(1) longitude = float(samp['longitude'].values[0]) latitude = float(samp['latitude'].values[0]) housing_median_age = float(samp['housing_median_age'].values[0]) total_rooms = float(samp['total_rooms'].values[0]) total_bedrooms = float(samp['total_bedrooms'].values[0]) population = float(samp['population'].values[0]) households = float(samp['households'].values[0]) median_income = float(samp['median_income'].values[0]) sample_data.append([longitude,latitude,housing_median_age,total_rooms,total_bedrooms,population,households,median_income]) return sample_data def predict_price(longitude,latitude,housing_median_age,total_rooms,total_bedrooms,population,households,median_income): # initialize data of lists. data = {'longitude':float(longitude), 'latitude':float(latitude), 'housing_median_age':float(housing_median_age), 'total_rooms':float(total_rooms), 'total_bedrooms':float(total_bedrooms), 'population':float(population), 'households':float(households), 'median_income':float(median_income), } test_features = pd.DataFrame(columns=['longitude', 'latitude', 'housing_median_age', 'total_rooms', 'total_bedrooms', 'population', 'households', 'median_income']) # Create DataFrame test_features = test_features.append(data,ignore_index=True) test_features = test_features.dropna(subset=["total_bedrooms"]) ## 3. 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 pred = lin_reg.predict(test_features_normalized)[0] map_file = draw_map_customize(longitude,latitude, fig_id='test',fig_extension='png' ) return pred,map_file ### configure inputs/outputs set_longitude = gr.inputs.Slider(-124.350000, -114.310000, step=0.5, default=-120, label = 'Longitude') set_latitude = gr.inputs.Slider(32, 41, step=0.5, default=33, label = 'Latitude') set_housing_median_age = gr.inputs.Slider(1, 52, step=1, default=10, label = 'Housing_median_age (Year)') set_total_rooms = gr.inputs.Slider(1, 40000, step=5, default=10000, label = 'Total_rooms') set_total_bedrooms = gr.inputs.Slider(1, 6445, step=5, default=5000, label = 'Total_bedrooms') set_population = gr.inputs.Slider(3, 35682, step=5, default=10, label = 'Population') set_households = gr.inputs.Slider(1, 6082, step=5, default=10, label = 'Households') set_median_income = gr.inputs.Slider(0, 15, step=0.5, default=10, label = 'Median_income') set_label = gr.outputs.Textbox(label="Predicted Housing Prices") # define output as the single class text set_out_images = gr.outputs.Image(label="Visualize your location") ### configure gradio, detailed can be found at https://www.gradio.app/docs/#i_slider interface = gr.Interface(fn=predict_price, inputs=[set_longitude, set_latitude,set_housing_median_age,set_total_rooms,set_total_bedrooms,set_population,set_households,set_median_income], outputs=[set_label,set_out_images], examples_per_page = 2, examples = get_sample_data(10), title="CSCI4750/5750 Demo 3: Web Application for Housing Price Prediction", description= "Click examples below for a quick demo", theme = 'huggingface', layout = 'vertical' ) interface.launch(debug=True)