End-to-End ML Project

Student Exam Performance Prediction - End-to-End Machine Learning Project

Overview

This project implements a complete end-to-end machine learning pipeline to predict a student's mathematics score based on multiple demographic and academic indicators. The goal is to identify the most influential factors affecting student performance and to provide a deployable predictive system accessible through a web interface.

The solution is structured for production-grade scalability, maintainability, and modularity, following standard ML engineering best practices.

---

Objectives

1. Build a robust regression model that predicts math scores from student characteristics and prior academic data.
2. Establish a modular and reusable machine learning pipeline for training and inference.
3. Deploy the model using Flask, enabling real-time predictions via a web-based form.
4. Ensure maintainable code design with proper logging, exception handling, and configuration management.

---

Methodology

The project workflow consists of the following stages:

1. Data Ingestion

• Load the Student Performance dataset (CSV format).
• Handle missing or inconsistent data if present.
• Split the dataset into training and test subsets.

2. Data Transformation

• Encode categorical variables (e.g., gender, ethnicity, parental education).
• Scale numerical features using standardization techniques.
• Build and persist the preprocessing pipeline using pickle for future inference.

3. Model Training

• Train multiple regression algorithms such as Linear Regression, Random Forest, and Gradient Boosting.
• Evaluate models using Mean Absolute Error (MAE), Mean Squared Error (MSE), and R² score.
• Select the best-performing model and persist it for deployment.

4. Model Evaluation

• Evaluate model performance on unseen test data.
• Analyze prediction errors and bias.
• Validate the pipeline's reproducibility and stability.

5. Deployment

• Integrate the trained model and preprocessor into a Flask application.
• Accept user input through an HTML interface for real-time prediction.
• Render predicted math scores dynamically on the web page.

---

Key Features

• End-to-End ML Pipeline: From raw data ingestion to live model inference.
• Modular Design: Each stage (data processing, model training, prediction) is encapsulated in separate classes and scripts.
• Exception Handling and Logging: Centralized error management using custom exception and logging modules.
• Web Deployment: Interactive prediction via Flask-based web interface.
• Scalable Architecture: Supports integration with CI/CD and cloud deployment workflows.

---

Technologies Used

• Programming Language: Python 3.10+
• Machine Learning: scikit-learn, pandas, numpy
• Model Deployment: Flask
• Data Serialization: pickle
• Web Frontend: HTML5, Bootstrap 5
• Development Tools: Git, VS Code, Jupyter Notebook

---

Installation and Setup

Prerequisites

Ensure the following are installed on your system:

• Python 3.10 or higher
• pip package manager
• Git

Setup Steps

1. Clone the repository:

   git clone https://github.com/<your-username>/student-performance-prediction.git
   cd "End-to-End ML Project"

2. Create and activate a virtual environment:

python3 -m venv venv
source venv/bin/activate      # Mac/Linux
venv\Scripts\activate         # Windows

3. Install dependencies:

pip install -r requirements.txt

4. Run the Flask application:

python3 app.py

5. Open the web application in your browser:

http://127.0.0.1:5000/