Performance Comparison of XGBoost and CatBoost Algorithm in Credit Card Fraud Detection with Streamlit-Based Web Application

Credit cards have transformed from a convenient payment method to a symbol of financial freedom and accessibility, yet their increased usage has also paralleled a rise in fraudulent activities, driven by sophisticated techniques. Credit card fraud, a significant cybercrime, involves unauthorized access to sensitive card information for fraudulent transactions, causing substantial financial losses. This project aims to compare machine learning algorithms specifically CatBoost and XGBoost for effective fraud detection. The proj ect is done employing a simulated credit card dataset, and it compares algorithm in terms of accuracy, precision, fl-score, roc-auc in order to find the best model. Both the model achieved the highest accuracy and precision. The goal is to enhance the capacity of fraud detection by employing models for two-step verification so that differentiation between fraud and legitimate transaction could be swift. It uses the streamlit framework to develop an accessible machine learning web application, thus making the suggested models practical for application towards enhancing credit card security.