Deploying a Machine Learning Model Using Flask

Overview

• Discussion on deploying a machine learning model using Flask.
• Apologies for the delay in the presentation.
• Focus on simplicity in deployment techniques.

Required Components

1. Model Building File (Python file)

   • Responsible for creating the model, feature engineering, and data pre-processing.
   • Example model: Linear regression.

2. Index.html File

   • Acts as the front-end web app for user input.
   • Interacts with the model's API to produce output.

3. Additional Python File (app.py)

   • Contains Flask code to create APIs.
   • APIs:
     • /reading (example)
     • /predict (to get predictions based on user input)

Dataset

• Dataset used: hiring.csv
• Fields: Experience, Test Score, Interview Score, Salary
• Independent Features: Experience, Test Score, Interview Score
• Dependent Feature: Salary

Data Pre-Processing Steps

• Replace missing experience data with 0 (no experience).
• Handle NaN values in test scores by replacing with the mean.
• Convert strings to integers for experience data.
• Test and interview scores range from 0 to 10.

Feature Engineering

• Convert experience strings to integers.
• Replace missing values with mean values for test scores and interview scores.
• Independent features are prepared for linear regression.

Model Creation

• Apply linear regression using the processed dataset.
• Save the model using the pickle library (creates a .pkl file).
• Ensure the model file is ready for deployment.

Creating Flask Environment

• Import necessary libraries: numpy, pickle, flask.
• Initialize Flask app.
• Load the pre-trained model using pickle.

Flask Routes

1. Home Route (/)

   • Redirects to the index.html file.

2. Predict Route (/predict)

   • Accepts POST requests with inputs for prediction.
   • Uses input values to generate predictions through the model.
   • Outputs the predicted salary back into the index.html file.

3. Hard-coded JSON API

   • Example API for testing with hard-coded values.

Running the Application

• Structure the directories: static files, templates, and the main Python app file.
• Use command prompt to run the Flask app: python app.py.
• Access the web app via a browser at the specified URL.

User Interaction

• Input fields for:
  • Years of Experience
  • Test Score (0-10)
  • Interview Score (0-10)
• After submitting, the predicted salary is displayed.

Conclusion

• Reminder to handle input properly to ensure accurate predictions.
• Encouragement to subscribe for more content
• Emphasis on continuing to learn and improve.
• Apologies for previous delays in content.