With the exponential rise in online financial transactions, credit card fraud has become a pressing challenge for both consumers and financial institutions. Conventional rule-based detection systems are increasingly ineffective in identifying sophisticated and evolving fraud patterns, often resulting in high false positive rates and delayed responses. This project proposes a machine learning–based fraud detection framework designed to enhance real-time accuracy, scalability, and adaptability. The methodology involves preprocessing real-world credit card transaction datasets, addressing data imbalance through techniques such as the Synthetic Minority Over-sampling Technique (SMOTE), and training multiple classification algorithms including Logistic Regression, Decision Tree, Random Forest, and XGBoost. The models are evaluated using performance metrics such as precision, recall, F1-score, and ROC-AUC, ensuring a balanced approach to fraud detection. Furthermore, the system integrates a Streamlit-based interactive interface that enables real-time transaction analysis and user-friendly fraud prediction. Experimental results highlight the effectiveness of the proposed system in minimizing false alarms while maintaining high detection accuracy. This research establishes a scalable and practical solution for combating credit card fraud, with promising applications in financial institutions, e-commerce platforms, and payment gateways.
Khaja Mahabubullah (Mon,) studied this question.