📊 Student Performance Predictor with MLOps Deployment

This project predicts student academic performance using socio-economic and educational attributes. It integrates Machine Learning modeling, interactive UI development with Flask, and MLOps deployment on AWS with CI/CD automation using GitHub Actions and Docker.

---

🚀 Features

• Predictive Modeling: Achieves ~90% R² score using regression models on student test data.

• Data Analysis: In-depth data cleaning, preprocessing, and EDA using Pandas, Matplotlib, and Seaborn.

• ML Models Used:

  • Linear Regression, Ridge, Lasso
  • Random Forest, SVM, KNN, Decision Tree
  • XGBoost, CatBoost, AdaBoost

• Best Performing Models:

  • Ridge Regression: 0.8805
  • Linear Regression: 0.8803
  • CatBoost Regressor: 0.8516

---

🧠 Tech Stack

Component	Technology
Language	Python
ML Libraries	Scikit-learn, XGBoost, CatBoost
Web Framework	Flask
Visualization	Matplotlib, Seaborn
Containerization	Docker
Cloud	AWS EC2, AWS ECR
CI/CD	GitHub Actions
UI	HTML Templates, Streamlit

---

📂 Project Structure

├── app.py                  # Flask entry point
├── Dockerfile              # Docker setup
├── notebooks/              # EDA & model training
├── templates/              # HTML files
├── src/
│   ├── pipeline/           # Custom prediction pipeline
│   ├── components/         # Data ingestion, transformation, training
├── artifacts/              # Stored models and scalers
├── .github/workflows/      # CI/CD workflow definitions
├── requirements.txt        # Dependencies
└── setup.py                # Project packaging

---

🧪 How to Run Locally

# 1. Clone the repository
git clone https://github.com/vaibhavbunny/mlproject-aws-CI-CD-Projects.git
cd mlproject-aws-CI-CD-Projects

# 2. Create a virtual environment & activate
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# 3. Install dependencies
pip install -r requirements.txt

# 4. Run Flask app
python app.py

Visit http://localhost:5000 to use the web app.

---

🐳 Run with Docker

docker build -t student-performance-app .
docker run -p 5000:5000 student-performance-app

---

☁️ Deployment

• Deployed on AWS EC2 with Docker
• Model artifacts stored on AWS ECR
• GitHub Actions automate training, testing, and deployment

---

📈 Results

Model	R² Score
Ridge Regression	0.8805
Linear Regression	0.8803
CatBoost Regressor	0.8516
AdaBoost Regressor	0.8498
Random Forest Regressor	0.8472
Lasso	0.8253
XGBRegressor	0.8216
K-Neighbors Regressor	0.7838
Decision Tree	0.7603