Credit Card Fraud Detection System Development

Aritra Mondal

ML Engineer

AI Engineer

Fullstack Engineer

Flask

JavaScript

XGBoost

Investment & Asset Management

Credit Card Fraud Detection System

Project Overview

This project is a simple, efficient, and scalable Credit Card Fraud Detection System that predicts the likelihood of a transaction being fraudulent based on given transaction data. Using a trained model, the system evaluates each transaction’s attributes and classifies it as either potentially fraudulent or safe.

Technologies Used

Below are the primary technologies and tools used to develop this project:

Technology Description Logo Python The core programming language used for building the backend, model training, and handling data.

Flask A lightweight web framework in Python, used for creating the backend REST API.

Random Forest & XGBoost Machine learning models used for fraud detection. Random Forest provides high accuracy for initial classification, and XGBoost helps fine-tune and optimize the model further.

HTML/CSS/JavaScript These are used to create a simple and clean user interface for accepting user inputs and displaying prediction results.

GitHub Used for version control and hosting the code repository for collaborative development.

Project Structure

Credit Card Fraud Detection/
├── backend/
│   ├── app.py          
│   └── services/
│       └── model_service.py
│   ├── api/
│       └── routes.py
│   ├── frontend/
│       └── index.html                
│       └── styles.css                
│       └── script.js
│   └── models/
│       └── ensemble_model.pkl
│       └── scaler.pkl
│       └── threshold.pkl
├── datasets/
│    └── creditcard.csv       #Dataset
├── models/
│    └── model_train.py
│    └── preprocess.py
├── requirements.txt
├── README.md

Model Training Steps

The machine learning model in this project was trained using the Credit Card Fraud Detection Dataset, which contains anonymized transaction data labeled as either fraudulent or legitimate. Below are the key steps to train the model:

Data Loading The data (creditcard.csv) is loaded and preprocessed for the model training pipeline.

Data Preprocessing

Check for any missing values and handle them appropriately.

Normalize or standardize features to bring them to a comparable scale.

Split the dataset into training and testing sets.

Model Selection

Random Forest and XGBoost classifiers were selected due to their effectiveness in handling complex datasets and providing high accuracy in classification tasks.

The model is trained using the training dataset, and hyperparameter tuning is done to achieve optimal performance.

Model Evaluation

After training, the model is evaluated using the testing set.

Key metrics such as accuracy, confusion matrix, and classification report (precision, recall, F1-score) are used to validate the model’s effectiveness.

Model Saving

The trained model is saved as a .pkl file for easy loading and prediction in the Flask backend.

Running the Model Training Script

The model_train.py script handles the entire training process. To run the model training, execute the following command:

python model_train.py

Backend API Endpoints

Base Endpoint (/) This endpoint serves as a welcome message to confirm the API is running.

Prediction Endpoint (/predict) This endpoint accepts JSON data with features Time, V1 to V28, and Amount. The model returns a prediction along with a probability score indicating the likelihood of the transaction being fraudulent.

Example JSON format:

{
    "Time": 406,
    "V1": -2.3122265423263,
    "V2": 1.95199201064158,
    ...
    "V28": -0.143275874698919,
    "Amount": 239.93
}

Response:

{
    "prediction": 1,
    "probability": 0.85
}

Frontend UI

The frontend consists of three primary files in the /frontend folder:

index.html Provides a user-friendly form to input individual transaction features or paste them collectively. It includes a button to parse comma-separated values into JSON format, which is then sent to the backend for prediction.

script.js JavaScript logic for handling form submission, parsing collective data, and displaying the prediction result.

styles.css Basic CSS styling for a simple and clean user interface.

How to Run the Project

1. Clone the Repository

git clone https://github.com/we-re-wolf/Creditcard-Fraud-Detection
cd project-directory

2. Install Dependencies

pip install -r requirements.txt

3. Run the Backend Server

Start the Flask app with:

cd backend
python app.py

The server will run at http://127.0.0.1:5000/.

4. Access the Frontend

Open frontend/index.html in a web browser to interact with the UI.

Testing the System

Run the Server Make sure the backend Flask server is running and ready to accept requests.

Submit Test Data Use the frontend form to submit test data. Either enter each feature individually or paste a comma-separated list of values.

View Prediction After submission, the prediction message will indicate whether there is a high likelihood of fraud based on the model’s evaluation.

Future Enhancements

Potential improvements include:

Enhancing the UI/UX: Incorporate advanced animations and dynamic feedback.

Model Upgrades: Experiment with additional models or ensemble techniques to further improve prediction accuracy.

Feature Engineering: Explore additional features or external data that may help improve model performance.

Like this project

Posted Jul 6, 2025

Developed a scalable Credit Card Fraud Detection System using machine learning models.

Likes

Views