Credit Card Fraud Detection Project

Table of Contents

1. Project Overview
2. Dataset
3. Installation
4. Results
5. Dependencies
6. Scope for Improvement
7. License

• This project focuses on detecting fraudulent credit card transactions using machine learning techniques.
• By analyzing transaction data, the goal is to build a robust model that can accurately identify fraudulent activities while minimizing false positives.

Project Overview

Credit card fraud is a significant issue in the financial industry. This project aims to address this problem by building a machine learning model capable of identifying fraudulent transactions. The project emphasizes:

• Handling severe class imbalance.
• Feature selection and importance analysis.
• Model evaluation and hyperparameter tuning.

Dataset

The dataset contains anonymized credit card transaction data, including:

• Features: 30 numerical features (V1-V28, Time, Amount).
• Target: Class (0 = Normal, 1 = Fraud).

Getting Started

Prerequisites

• Python 3.9 or higher

Installation

1. Clone the repository:

   git clone https://github.com/kavin/cc-fraud-detection.git
   cd cc-fraud-detection

2. Install dependencies:

   pip install -r requirements.txt

3. Run the Jupyter Notebook:

   jupyter notebook cc_fraud_detection.ipynb

Results

• Balanced Dataset: Achieved a 1:1 ratio using SMOTE.
• Model Performance:
  • High precision and recall for fraud detection.
  • Improved accuracy after hyperparameter tuning.
• Feature Importance: Identified key features contributing to fraud detection.

Dependencies

Category	Dependency
Analysis	numpy, pandas
Visualization	matplotlib, seaborn
Preprocessing	imblearn (SMOTE)
Model Inference	xgboost
Model Selection	scikit-learn (GridSearchCV)
Model Evaluation	scikit-learn (accuracy_score, classification_report)

Scope for Improvement

• Explore additional models like Random Forest and Neural Networks.
• Implement real-time fraud detection pipelines.

License

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