Credit Card Fraud Detection

A machine learning-powered web application designed for banks to detect credit card fraud in real-time. This solution enables banks to process bulk transaction data, monitor real-time fraud activities, and utilize actionable insights to mitigate risks associated with fraudulent transactions.

Features

• File Upload: Allows users to upload a CSV file containing multiple transactions for batch processing.
• Real-Time Monitoring: Displays incoming transactions and their fraud status in real-time.
• Progress Bar: Shows upload progress during file processing.
• Filtering: Use filters to focus on transactions of particular interest, such as those with high risk or exceeding a specific amount, making it easy to drill down into potential fraudulent activity.
• Summary Statistics: Analyze visualized data, including transaction distributions and fraud detection rates, to understand fraud patterns within the bank's transactions.
• Feedback Loop: Fraud analysts can mark transactions that have been incorrectly flagged, providing data for future model improvement and making the system more adaptive to new fraud tactics.

Tech Stack

• Frontend: React, Axios, HTML, CSS
• Backend: Flask, Flask-CORS, Pandas, Joblib
• Model: Machine learning model (Decision Trees, MLP Classifier, Ramdom Forest Model, Neural Network, Gradient Boosting) trained for fraud detection using a transaction dataset.
• Additional Libraries: Chart.js or Recharts for graphs and visualizations (frontend).

Getting Started

Prerequisites

• Node.js and npm installed on your machine.
• Python 3 and pip installed.
• Recommended Python libraries: flask, flask-cors, pandas, joblib, sklearn.

Installation

1. Clone the Repository

git clone https://github.com/Danielmark001/fraud-detection-ml.git
cd fraud-detection-ml

Running the Application

1. Start the Backend (Flask)

python app.py

The backend will run on http://localhost:5000.

2. Start the Frontend (React) In a new terminal window, start the React development server:

npm start

The frontend will run on http://localhost:3000.

User Guide

1. Upload Transactions for Batch Processing

Go to the Upload Transactions section in the app. Click the "Browse" button to select a CSV file. A sample template is available for download to format your data correctly. Click the "Upload" button. A progress bar will indicate the upload status in real-time.

2. Real-Time Monitoring

View incoming transactions in the Real-Time Monitoring section. Apply filters (amount, risk level, date) to narrow down the displayed transactions. View summary statistics with interactive charts showing transaction distribution and fraud detection rates.

3. Customizing Detection Settings

Adjust model sensitivity in the settings section (if available). Use the feedback loop to mark incorrect predictions, helping to improve the model in future versions.

Sample CSV Template

A sample CSV template (sample-template.csv) is provided in the frontend/public directory. Users can download and format their transaction data accordingly for batch processing.

API Endpoints

• POST /upload: Uploads a CSV file for batch processing.
• GET /transactions: Retrieves the latest transactions for real-time monitoring.
• POST /feedback: Allows users to submit feedback on the model's predictions for future improvements.

Future Enhancements

• Integration with Bank Systems: Enable integration with the bank's transaction processing systems for real-time fraud monitoring.
• Enhanced Authentication: Add user roles and authentication to control access based on user responsibilities within the bank.
• Data Storage: Introduce database support to store transaction history and feedback securely.

Dataset Source

Axi, E. (2018). Synthetic Financial Datasets For Fraud Detection [Data set]. Kaggle. https://www.kaggle.com/datasets/ealaxi/paysim1/data

Ward, J. (2024). Metaverse Financial Transactions Dataset. https://www.kaggle.com/datasets/faizaniftikharjanjua/metaverse-financial-transactions-dataset