Real-time Fraud Detection System

A machine learning system for real-time credit card fraud detection using MLflow, FastAPI, Kafka, and Streamlit.

Prerequisites

• Docker and Docker Compose
• Python 3.9+
• Git

Installation

1. Clone the repository:

git clone git@github.com:Thejaswi05/realtime_fraud_detection.git

cd realtime-fraud-detection

2. Create and activate a virtual environment (optional but recommended):

```bash
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

Running the System

1. Start all services:

# Clean start
docker-compose down -v
docker-compose up -d

2. Train the model:

docker-compose run --rm train

3. Access the services:

• MLflow UI: http://localhost:5001
• Model Playground: http://localhost:8501
• API Documentation: http://localhost:8000/docs

Using the Streamlit Playground

The Streamlit playground provides an interactive interface to test the fraud detection model.

1. Access the playground at http://localhost:8501

2. Input transaction details:

   • Amount: Enter the transaction amount in dollars
   • Time: Select the transaction's position in the sequence

3. Click "Check for Fraud" to get predictions

4. Understand the results:

   • Prediction: Shows if the transaction is legitimate or fraudulent
   • Confidence: Shows the model's confidence in its prediction
   • Feature Analysis: Displays the top 10 influential features
     • Blue bars (positive values): Transaction patterns the model has learned
     • Red bars (negative values): Different transaction patterns
     • Bar length: Strength of the feature's influence
     • The combination of all features determines the final prediction

5. Feature Interpretation:

   • Time: Position in transaction sequence (helps detect patterns)
   • Amount: Transaction value
   • V1-V28: Anonymized transaction characteristics

Development

To modify the system:

1. Update model training:

3. Update playground:

```bash
docker-compose restart playground

Troubleshooting

1. Port conflicts:

# Stop all containers and remove volumes
docker-compose down -v
# Check for running containers
docker ps
# Kill specific ports if needed
sudo lsof -i :5001  # Check MLflow port
sudo lsof -i :8501  # Check Streamlit port

2. MLflow connection issues:

# Restart MLflow
docker-compose restart mlflow
# Check logs
docker-compose logs mlflow

3. Model loading errors:

# Ensure model is trained
docker-compose run --rm train
# Check MLflow UI for successful runs

Contributing

1. Fork the repository
2. Create a feature branch
3. Commit changes
4. Push to the branch
5. Create a Pull Request

This README:

1. Provides clear installation steps
2. Explains how to run each component
3. Details the Streamlit playground usage
4. Includes troubleshooting guides
5. Explains project structure