Name: Shubham Verma
Roll Number: G24AI1002
Course: MLOps
Instructor: Dr. Pratik Mazumder

---

📌 Project Overview

This repository implements a complete MLOps pipeline using LinearRegression from scikit-learn on the California Housing dataset. It includes:

• Model training
• Manual quantization (8-bit)
• Inference and evaluation
• Docker containerization
• CI/CD integration with GitHub Actions
• Automated testing

---

📂 Repository Structure

.
├── src/
│   ├── train.py           # Train Linear Regression model
│   ├── quantize.py        # Quantize model coefficients
│   ├── predict.py         # Make predictions
│   └── utils.py           # Shared functions
├── tests/
│   └── test_train.py      # Unit tests using pytest
├── .github/
│   └── workflows/
│       └── ci.yml         # GitHub Actions workflow
├── model.joblib           # Pretrained model (generated via train.py)
├── requirements.txt       # Project dependencies
├── Dockerfile             # Container setup
├── .gitignore             # Ignore cache/artifact files
└── README.md              # Project documentation

---

🚀 How to Run

1. Setup Environment

python -m venv venv
venv\Scripts\activate        # On Windows
pip install -r requirements.txt

2. Train the Model

python src/train.py

3. Quantize the Model

python src/quantize.py

4. Predict

python src/predict.py

5. Run Unit Tests

pytest

6. Build & Run with Docker

docker build -t housing_model .
docker run --rm housing_model

---

🧪 CI/CD Workflow

GitHub Actions runs on every push to main, consisting of:

1. ✅ test suite: runs unit tests using pytest
2. ✅ train and quantize: trains model and performs quantization
3. ✅ build and test container: builds Docker image and verifies predictions

All jobs must pass for the workflow to succeed.

---

📷 Output

• R2 Score & MSE

• Sample Vs Actual Prediction

---

🔐 Notes & Constraints

• Only LinearRegression is used.
• No hardcoded values.
• Code is modularized and organized.
• Only the main branch exists.
• CI/CD is implemented using GitHub Actions.

---