simplelinearregress 0.0.1

This is a simple implementation of linear regression

Simple Linear Regression

This is a simple implementation of linear regression.

1. Install package simplelinearregress package by pip. All dependent packages have been already installed.

• pip install simplelinearregress ==0.0.1

2. Train model with diabetes dataset, evaluate the trained model, save the trained model to disk

   import os
   import pickle
   from linear_reg.simple_linear_regr import SimpleLinearRegression
   from linear_reg.simple_linear_regr_utils import generate_data, evaluate
   
   if __name__ == "__main__":
       # load diabetes data
       X_train, y_train, X_test, y_test = generate_data()
   
       # load model
       model = SimpleLinearRegression()
   
       # trained model
       model.fit(X_train, y_train)
   
       # evaluation model
       predicted = model.predict(X_test)
       evaluate(model, X_test, y_test, predicted)
   
       # save trained model
       if not os.path.exists('saved_model'):
           os.makedirs('saved_model')
       model_path = os.path.join("saved_model", "linear_model.dat")
   
       with open(model_path, 'wb') as saved_model:
           pickle.dump(model, saved_model)
   

3. The evaluated result is as follows:

   Slope: [[937.18913157]]; Intercept: [152.9193589]
   Mean squared error: 2549.27
   Coefficient of determination: 0.47
   ****** Success ******
   

4. Create a RESTful API server with Flask

• Create a configure file in configs folder, named .env. This file includes all configurations of servers.

  # ip and port of server
  HOST=0.0.0.0
  PORT=8080
  MODEL_DIR= "saved_model"
  MODEL_NAME="linear_model.dat"
  API_KEYS=I1TIISEUKJBTJVNO3M24UC7L71CAW3T0,AY814XUOHKNI4R0DIPXYRRO07L5EIQFX,REEOBZZ6FL03VDIAX6LR71RFNFIJYTJJ,FRY7XKAGFZG3WGNX9V4MK7A7O6T8VORK,AU39QYVZIKCUITCOYMZ6STUDC6R9GBKF
  

• Run API server. Note that you must import SimpleLinearRegression

  from linear_reg.main import main
  from linear_reg.simple_linear_regr import SimpleLinearRegression
  
  if __name__ == '__main__':
      main()
  

• Server ready started as follows:

   * Serving Flask app 'linear_reg.main'
   * Debug mode: on
  WARNING: This is a development server. Do not use it in a production deployment. Use a production WSGI server instead.
   * Running on all addresses (0.0.0.0)
   * Running on http://127.0.0.1:8080
   * Running on http://192.168.10.102:8080
  Press CTRL+C to quit
   * Restarting with stat
   * Debugger is active!
   * Debugger PIN: 236-046-492
  

5. Create a client to request RESTful API server

   import requests
   import ast
   from linear_reg.simple_linear_regr_utils import generate_data
   
   if __name__ == '__main__':
       # load data
       X_train, y_train, X_test, y_test = generate_data()
   
       api_key = "I1TIISEUKJBTJVNO3M24UC7L71CAW3T0"
   
       # Header input
       headers = {"Content-Type": "application/json", "regression-api-key": api_key}
   
       # API url
       url = 'http://localhost:8080/batch'
   
       # sending data
       data = {'X_test': X_test.tolist()}
   
       # Request server
       response = requests.post(url, json=data, headers=headers)
   
       # Convert server response into JSON format
       response_content = response.content
       response_content = response.content.decode("utf-8")
       response_content = ast.literal_eval(response_content)
       print("Predict results:")
       print(response_content["results"])
   

• The response prediction is as follows:

  Predict results:
  [225.89220475078582, 115.78961465717714, 163.26504341313685, 114.77949915173122, 120.8401921844069, 158.2144658859071, 235.99335980524535, 121.85030768985287, 99.62776657004196, 123.87053870074476, 204.67977913642085, 96.5974200537041, 154.1740038641233, 130.9413472388664, 83.46591848290672, 171.34596745670444, 138.01215577698807, 138.01215577698807, 189.5280465547316, 84.47603398835268]