SECURING DIGITAL TRANSACTIONS: MACHINE LEARNING FRAMEWORKS FOR FRAUD DETECTION IN PAYMENT SYSTEMS

Abstract

The widespread adoption of digital payment systems has revolutionized financial transactions, offering seamless and efficient services globally. However, this advancement has heightened cybersecurity risks, with fraudulent transactions posing a significant threat. These fraudulent activities result in financial losses and undermine trust in digital platforms, necessitating robust detection mechanisms. This paper proposes a machine learning (ML)-based framework for detecting fraudulent transactions in online payment systems. Seven models, including Logistic Regression (LR), K-Nearest Neighbors (KNN), Decision Trees (DT), Random Forest (RF), Extreme Gradient Boosting (XGB), Naive Bayes (NB), and Support Vector Classifier (SVC), were evaluated. To address the challenge of imbalanced datasets, over- sampling and under sampling techniques were applied, enhancing model sensitivity to minority-class samples. Experimental results revealed that ensemble models, particularly RF and XGB, achieved the highest accuracy of 99.3% and 99.4%, respectively. These models outperformed simpler classifiers such as LR and NB in key metrics, including precision, recall, and F1-score. This research highlights the potential of ML to strengthen cybersecurity in digital payment systems. By addressing challenges such as data imbalance and scalability, the proposed framework provides actionable insights for developing next-generation fraud detection systems, enhancing trust and security in the digital economy.

Index Terms—Fraud Detection, ML, Cybersecurity, Over- sampling, Under sampling, Ensemble Models, Digital Payment Systems.