Fraud Detection in Banking Transactions Using Artificial Intelligence-Based Analysis of Database Patterns

Fraudulent activities in banking systems result in significant financial losses annually. In this study, we propose a hybrid artificial intelligence approach using decision trees, random forest classifiers, and deep learning models to detect anomalies and fraud in banking transaction databases. Our method incorporates preprocessing techniques such as data balancing (SMOTE), feature engineering, and time-based transaction analysis. Using a publicly available dataset, our system achieved an accuracy of ۹۸.۳% and an F۱-score of ۹۶.۷%, outperforming existing baseline models. These results suggest that AI-based analysis of transactional databases can provide an efficient tool for real-time fraud detection.

کلیدواژه ها:

Fraud Detection ، Banking Transactions ، Machine Learning ، Database Analysis ، Deep Learning ، Random Forest ، SMOTE