TrustCheck:Secured Banking Fraud Detection Using CatBoost and XGBoost

Modern-day online banks need effective fraud detection platforms because of increased cyber threats in current times. The current methods lack success in detecting recent fraud strategies so money continues to disappear. The project demands the combination of CatBoost and XGBoost machine learning algorithms to implement quick and efficient fraud detection operations. The accuracy improvement steps in the preprocessing stage incorporate standardization together with feature engineering and then balance the data. The performance of your model becomes better after conducting Bayesian hyperparameter tuning. The system utilizes precision and recall together with F1score and ROC-AUC score metrics for its assessment. Real-time alerts through the system notify users about uncertain financial operations. A built-in automatic adaptive feature enables the system to identify new fraud patterns which results in an adaptable framework with robust operation and easy integration possibilities.