Credit Card Fraud Detection System

Project Overview

This project implements an advanced deep learning solution for detecting fraudulent credit card transactions. It combines the power of Recurrent Neural Networks (RNN) with multiple activation functions and a sophisticated architecture to achieve high-accuracy fraud detection.

Features

• Advanced RNN architecture with 6 hidden LSTM layers
• Multiple activation functions (ReLU, tanh, sigmoid, PReLU, ELU, SELU)
• Interactive web interface using Streamlit
• Support for both single and batch predictions
• Comprehensive visualization of results
• Detailed performance metrics and analysis

Technical Architecture

Deep Learning Model

• Input Layer: LSTM with 128 units
• Hidden Layers:
  1. LSTM (64 units) with ReLU activation
  2. LSTM (32 units) with tanh activation
  3. LSTM (16 units) with sigmoid activation
  4. LSTM (8 units) with PReLU activation
  5. LSTM (4 units) with ELU activation
  6. LSTM (2 units) with SELU activation
• Output Layer: Dense layer with sigmoid activation
• Regularization: Dropout layers (0.2) after each LSTM layer

Installation

1. Clone the repository:

git clone https://github.com/Mayankvlog/Credit-card-fraud-detection-Data-science-project.git
cd Credit-card-fraud-detection-Data-science-project

2. Create and activate virtual environment:

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

3. Install dependencies:

pip install -r requirements.txt

Usage

Running the Web Application

streamlit run app.py

Features of the Web Interface

1. Single Prediction:

   • Input transaction details manually
   • Get instant fraud probability predictions
   • Visual indicators for fraudulent/legitimate transactions

2. Batch Prediction:

   • Upload CSV files with multiple transactions
   • Get predictions for all transactions
   • Download results in CSV format
   • Visualize prediction distributions
   • View confusion matrix and performance metrics

3. Model Information:

   • View model architecture
   • Access performance metrics
   • Read usage instructions

Data Format

The model expects the following features:

• Time: Seconds elapsed between transactions
• V1-V28: Principal components from PCA
• Amount: Transaction amount

Model Performance

• Uses binary cross-entropy loss
• Implements early stopping to prevent overfitting
• Includes comprehensive evaluation metrics:
  • ROC curve and AUC score
  • Precision-Recall curve
  • Confusion matrix
  • Classification report

Project Structure

├── app.py                   # Streamlit web application
├── requirements.txt         # Python dependencies for model
├── requirements_app.txt     # Dependencies for web app
├── data/
│   └── creditcard.csv      # Dataset file
├── model/
│   ├── credit_card_fraud_model.h5  # Saved model
│   └── scaler.pkl          # Saved scaler
└── credit_card_fraud_detection.ipynb  # Model development notebook

Contributing

Contributions are welcome! Please feel free to submit pull requests.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Dataset: Credit Card Fraud Detection dataset
• Libraries: TensorFlow, Keras, Streamlit, scikit-learn
• Visualization: Matplotlib, Seaborn

Contact

• Author: Mayank
• GitHub: @Mayankvlog