Fraud Detection in Digital Payment Systems using Machine Learning

This study developed and compared four machine learning models for fraud detection in digital payment transactions. XGBoost achieved the highest accuracy, but Logistic Regression achieved the highest recall and was selected as the preferred model, showing that accuracy alone is not an adequate criterion for imbalanced fraud detection tasks - recall and the precision-recall balance matter more when the cost of missing fraud is high. SHAP analysis improved the interpretability of the selected model, and the resulting risk-scoring framework and early-warning alert system make the output directly usable in a real payment monitoring workflow. The low precision and ROC-AUC values obtained across all models indicate that the models are not yet ready for production use; this is attributable to the modest dataset size, its class imbalance, and the absence of richer behavioural or sequential features. Future work should explore larger and more diverse transaction datasets, behavioural and temporal features, and more advanced algorithms to build a reliable, real-time fraud detection system for digital payments.

1. Mahdi, M. D. H., Rezaul, K. M., Rahman, M. A.: Credit Fraud Detection in the Banking Sector in UK: A Focus on E-Business. Int. Conf. on Innovative Computing Technology (2010) 2. Al-Khatib, A. M.: Payment Fraud Detection Techniques. International Journal of Computer Applications (2012) 3. Franca Lima, R., Machado Pereira, A. C.: Feature Selection Approaches to Fraud Detection in E-Payment Systems. E-Commerce and Web Technologies (2015) 4. Roy, A., Sun, J., Mahoney, R., Alonzi, L., Adams, S., Beling, P.: Deep Learning Detecting Fraud in Credit Card Transactions. Systems and Information Engineering Design Symposium (2018) 5. Diadiushkin, A., Sandkuhl, K., Maiatin, A.: Fraud Detection in Instant Payments as Contribution to Digitalization in Banks. Int. Conf. on Business Informatics Research (2019) 6. Victoria Priscilla, C., Padma Prabha, D.: Influence of Optimizing XGBoost to Handle Class Imbalance in Credit Card Fraud Detection. Int. Conf. on Smart Electronics and Communication (2020) 7. Muhal, H., et al.: An Ensemble Approach for Credit Card Fraud Detection Using Champion-Challenger Analysis. Int. Conf. on Machine Learning and Data Engineering (2022) 8. Jain, N., Chaudhary, A., Kumar, A.: Credit Card Fraud Detection Using Machine Learning Techniques. Int. Conf. on Computing, Communication, and Intelligent Systems (2022) 9. Lande, J., Sharma, J.: Optimizing Logistic Regression for Fraud Detection in Digital Payments. International Journal of Computer Science and Engineering (2025) 10. Ebenezer, Mohan, S., Subalya, S., Jansi Rani, T.: AI-Powered Fraud Detection Framework for UPI and Digital Wallet Transactions Using Bi-LSTM with Attention. Int. Conf. on Artificial Intelligence and Smart Systems (2025) 11. Reserve Bank of India: Authentication Mechanisms for Digital Payment Transactions Directions, 2025. RBI (2025)