Credit Card Fraud Detection

A machine learning-based system designed to accurately detect fraudulent transactions using advanced algorithms and data preprocessing techniques.

---

Table of Contents

1. Introduction
2. Features
3. Technologies Used
4. Dataset
5. Installation
6. Usage
7. Results
8. Future Scope
9. License

---

Introduction

Credit card fraud poses a major challenge in the digital financial world. This project leverages machine learning algorithms to develop an efficient fraud detection system capable of identifying fraudulent transactions in real-time. The system addresses key challenges such as data imbalance and scalability.

---

Features

• Handles highly imbalanced datasets using SMOTE (Synthetic Minority Oversampling Technique).
• Implements advanced machine learning algorithms:
  • XGBoost, CatBoost, LightGBM, Random Forest, and Artificial Neural Networks (ANNs).
• Supports real-time fraud detection.
• Provides detailed performance evaluation metrics:
  • Accuracy
  • Precision
  • Recall
  • F1-Score
  • ROC-AUC
• Visualizes data insights with heatmaps, confusion matrices, and ROC curves.

---

Technologies Used

• Programming Language: Python
• Libraries:
  • Data Manipulation: pandas, numpy
  • Visualization: matplotlib, seaborn
  • Machine Learning: scikit-learn, imbalanced-learn, XGBoost, LightGBM, CatBoost
• Tools:
  • Jupyter Notebook
  • GitHub for version control

---

---

## **Dataset**

The dataset was sourced from [Kaggle](https://www.kaggle.com) and contains anonymized credit card transaction data. Key characteristics:
- **Transactions**: 284,807
- **Fraudulent Transactions**: 492 (~0.17%)
- **Features**: 30 anonymized features, including `Time` and `Amount`.

**Note**: Sensitive data is excluded for privacy.

---

## **Installation**

1. **Clone the Repository**:
   ```bash
   git clone https://github.com/your-username/credit-card-fraud-detection.git
   cd credit-card-fraud-detection

1. Create a Virtual Environment:

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

2. Install Dependencies:

   pip install -r requirements.txt
   

---

Usage

1. Preprocess the Data:

   • Clean, scale, and balance the dataset using src/preprocess.py.

2. Train the Models:

   • Run src/train_models.py to train multiple machine learning models.
   • Example: bash python src/train_models.py --model xgboost

3. Evaluate Models:

   • Evaluate models with src/evaluate.py to generate performance metrics and visualizations.

---

Results

Model Performance

Model	Accuracy	Precision	Recall	F1-Score	ROC-AUC
XGBoost	99.96%	95.0	82.0	88.0	0.999
CatBoost	99.96%	93.0	82.0	87.0	0.998
RandomForest	99.95%	89.0	81.0	85.0	0.997
ANN	99.95%	85.0	81.0	83.0	0.996

Visualizations

---

Future Scope

• Real-Time Processing: Implement streaming capabilities for live transaction data.
• Deep Learning: Explore architectures like LSTMs for sequential fraud detection.
• Global Deployment: Extend support for multi-currency and multi-lingual data.

---

License

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

---

Built With

• jupyter-notebook