END TO END ML PROJECT -Student Performance Prediction – ML Web App

This project is a complete end-to-end machine learning pipeline designed to predict a student's math score based on various inputs such as gender, ethnicity, parental education level, lunch type, test preparation course, reading score, and writing score. The project is deployed as a web application using Flask and hosted on Render.

Project Overview

The application takes input through a web form and returns the predicted math score using a trained regression model. It covers data ingestion, preprocessing, model training, and inference.

Features

• Clean modular structure (components separated logically)
• Data ingestion and train/test split
• Data transformation using pipelines with handling of categorical and numerical features
• Model training using multiple regressors (Random Forest, XGBoost, CatBoost, etc.)
• Automatic model selection based on best R² score
• Persisted preprocessor and model using dill
• Flask-based web interface for real-time prediction
• Deployed on Render

Project Structure

ML_PROJECTS/ ├── artifacts/ # Holds model.pkl, preprocessor.pkl, etc. ├── logs/ # Stores log files ├── notebook/ # Likely has your data CSV ├── src/ # Source files (components, pipeline, utils) ├── components/ # Data ingestion, transformation, training ├── pipeline/ # train_pipeline.py for full pipeline run ├── utils.py # Helper functions (save/load object, evaluation) ├── logger.py # Logging ├── exception.py # Custom exceptions ├── templates/ # HTML files for Flask ├── venv/ # Virtual environment (should be ignored) ├── application.py # Flask app ├── requirements.txt # contain all python dependencies ├── Procfile # For Render deployment,specifies how to run the Flask app using Gunicorn ├── .gitignore ├── README.md ├── setup.py # Can be used if you're packaging this as a module

How It Works

1. Data Ingestion: Loads CSV data from notebook/data/stud.csv and splits it into train/test sets.
2. Data Transformation: Applies preprocessing pipeline including SimpleImputer, StandardScaler, and OneHotEncoder with handle_unknown='ignore'.
3. Model Training: Trains multiple models, performs hyperparameter tuning, selects the best model, and saves it.
4. Prediction: Loads the model and preprocessor, transforms incoming form data, and returns the predicted math score.

How to Run Locally

1. Clone the repository: git clone https://github.com/DataProjectHub/ML_Students_Performance.git

2. Create a virtual environment: python -m venv venv venv\Scripts\activate

3. Install dependencies pip install -r requirements.txt

4. Run training pipeline (optional) python src/pipeline/train_pipeline.py

5. Start the Flask server python application.py

Visit http://localhost:5000/predictdata to use the form.

Deployment

The app is deployed on Render. You can access it here:
https://student-performance-deployment.onrender.com/predictdata

Tech Stack

• Python
• Scikit-learn
• XGBoost
• CatBoost
• Flask
• HTML/CSS (Bootstrap)
• Render (for deployment)

Credits

Project built and deployed by Pooja Anilkumar as part of learning full-cycle machine learning implementation and deployment.

Connect with me on LinkedIn
GitHub: DataProjectHub