Credit Card Fraud Detection — End-to-End MLOps Pipeline

This project is a production-ready MLOps pipeline built around a credit card fraud detection use case.
It demonstrates how to take a real-world imbalanced dataset, train a model, and deploy it using modern MLOps tools.

---

📌 Features

• Data Preprocessing & Engineering

  • Handles class imbalance with SMOTE oversampling.
  • Scales features with StandardScaler.
  • Splits data into train/test sets with stratification.

• Model Training

  • Fraud detection model using XGBoost.
  • Experiment tracking via MLflow.
  • Full evaluation with precision, recall, F1-score, ROC-AUC, PR-AUC.

• Deployment

  • REST API with FastAPI.
  • Containerized using Docker.
  • Ready for cloud deployment (AWS/GCP/Render/Heroku).

• MLOps Tools

  • MLflow for experiment tracking and model registry.
  • Configurable hyperparameters via CLI or YAML.
  • CI/CD ready (GitHub Actions / Docker Hub integration).

---

📝 Project Structure

mlops-churn/
│── data/               # (not included in repo) dataset storage        # Jupyter notebooks for EDA
│── src/
│   ├── app/
        └── main.py # main script       
│   └── train.py   # Training script with MLflow tracking
│── Dockerfile          # Container setup
│── docker-compose.yml  # Local testing with MLflow + API
│── requirements.txt    # Python dependencies
│── .gitignore          # Ignored files (datasets, venv, artifacts)
│── README.md           # Project documentation

---

⚙️ Setup & Installation

1. Clone the repository

git clone https://github.com/fafnirkyu/CardFraud.git
cd CardFraud

2. Create virtual environment

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

3. Install dependencies

pip install -r requirements.txt

---

🚀 Training the Model

From the SSH terminal (or VM console):

cd ~/CardFraud
source .env/bin/activate
python src/train.py

• Automatically tracks experiments with MLflow
• Logs metrics and model artifacts to mlruns/

Example Metrics (Credit Card Fraud Model)

Metric	Value
Accuracy	0.9995
Precision	0.8737
Recall	0.8469
F1-Score	0.8601
ROC AUC	0.9794
PR AUC	0.8746

Run MLflow UI locally:

mlflow ui --host 0.0.0.0 --port 5000

---

🌐 Running the API

Local (without Docker):

uvicorn src.api:app --reload --host 0.0.0.0 --port 8000

With Docker (locally or on AWS EC2):

docker-compose up --build -d

• FastAPI endpoint: http://<VM_or_EC2_IP>:8000/docs
• MLflow tracking UI: http://<VM_or_EC2_IP>:5000

✅ AWS Deployment Example:

• EC2 instance (free tier) with Docker installed
• docker-compose runs API + MLflow
• Accessible from your public EC2 IP for demos or resume showcase

---

📊 Example API Request

POST /predict
{
  "features": [0.1, -1.2, 0.34, ...]
}

Response:

{
  "fraud_probability": 0.87,
  "prediction": 1
}

---

🧑‍💻 Tech Stack

• Python 3.10
• FastAPI — API deployment
• XGBoost — fraud detection model
• MLflow — experiment tracking
• Docker — containerization
• AWS EC2 (optional) — cloud deployment
• GitHub Actions — CI/CD ready

---

📦 Deployment Options

• Local with Docker + FastAPI + MLflow
• Cloud / Resume-ready showcase:
  • AWS EC2 (Docker + docker-compose)
  • GCP Cloud Run / Vertex AI
  • Render / Railway / Heroku (quick deployment)

---

📌 Notes

• Dataset (creditcard.csv) is not included in the repo (too large).
  Download from Kaggle — Credit Card Fraud Detection.
  Place it inside data/creditcard.csv before training.
• .gitignore prevents committing datasets or model artifacts.

---

✨ Deliverable

End-to-End MLOps Pipeline:
Trained fraud detection model with:

• Accuracy: 0.9995
• Precision: 0.8737
• Recall: 0.8469
• F1-Score: 0.8601
• ROC-AUC: 0.9794
• PR-AUC: 0.8746

Deployed via FastAPI + Docker, tracked with MLflow, and production-ready for cloud deployment (AWS, GCP, or Render).

---

📜 License

MIT License — free to use and modify.