MLOps-pipeline

An end-to-end MLOps pipeline for predicting student academic risk (Graduate, Dropout, Enrolled). Features data versioning, experiment tracking (MLflow), hyperparameter tuning, FastAPI deployment, Docker containerization, and CI/CD automation with GitHub Actions.

Student Academic Risk Predictor: End-to-End MLOps Project

📌 Project Overview

This project implements a complete, production-grade machine learning pipeline to predict student academic risk in higher education. The model classifies students into three categories: Graduate, Dropout, or Enrolled.

It is built with a focus on MLOps best practices, demonstrating how to move from a raw dataset to a deployable, scalable API. The system includes automated training, hyperparameter tuning, experiment tracking, and containerized deployment.

Key Features

• Modular Codebase: Clean separation of concerns (data loading, preprocessing, training, tuning, deployment).
• Robust Preprocessing: Custom feature engineering and scikit-learn pipelines for data transformation.
• Experiment Tracking: Integration with MLflow to log parameters, metrics, and model artifacts.
• Hyperparameter Tuning: Automated optimization using RandomizedSearchCV.
• REST API: A high-performance FastAPI application for real-time predictions.
• Containerization: Fully Dockerized application for consistent deployment.
• CI/CD: Automated build and push workflows using GitHub Actions.

📂 Project Structure

student_risk_predictor/ 
├── .github/ 
│ └── workflows/ 
│ └── ci-cd.yml # GitHub Actions workflow for CI/CD 
├── app/ # FastAPI Application 
│ ├── init.py 
│ ├── main.py # API server logic 
│ └── schemas.py # Pydantic models for data validation 
├── artifacts/ # Generated files (models, encoders, metrics) 
│ └── (Populated automatically by scripts) 
├── data/ # Raw Data 
│ ├── train.csv # Training dataset 
│ └── test.csv # Test dataset (optional) 
├── mlruns/ # MLflow tracking data (auto-generated) 
├── notebooks/ # Jupyter Notebooks 
│ └── 1-Data-Exploration.ipynb 
├── src/ # Core ML Source Code 
│ ├── init.py 
│ ├── data_loader.py # Data loading and splitting logic 
│ ├── preprocessor.py # Preprocessing pipeline definition 
│ ├── train.py # Model training and selection script 
│ ├── tune.py # Hyperparameter tuning script 
│ └── utils.py # Helper functions 
├── .gitignore 
├── Dockerfile # Docker image configuration 
├── params.yaml # Configuration file for parameters 
├── requirements.txt # Python dependencies 
└── README.md # Project documentation

🚀 Getting Started

Prerequisites

• Python 3.8+
• Git
• Docker (optional for local dev, required for containerization)

Installation

1. Clone the repository:

   git clone [https://github.com/yourusername/student-risk-predictor.git](https://github.com/yourusername/student-risk-predictor.git)
   cd student-risk-predictor

2. Create and activate a virtual environment:

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

3. Install dependencies:

   pip install -r requirements.txt

4. Data Setup: Ensure you have the train.csv file placed inside the data/ directory.

🛠️ Usage Pipeline

Follow these steps to reproduce the entire training and deployment process.

1. Data Exploration (Optional)

Run the Jupyter notebook to understand the dataset distribution and correlations.

# Open the notebook in your editor or Jupyter Lab
notebooks/1-Data-Exploration.ipynb