Defending against digital thievery: a machine learning approach to predict e-payment fraud

The increased usage of credit cards has facilitated the development of e-commerce and electronic payment systems. However, this trend has also led to a surge in fraudulent activities. As a result, websites and e-commerce platforms that handle customer data have been required to establish efficient fraud prevention systems capable of detecting and preventing fraudulent electronic payment operations. Machine learning has emerged as a highly effective fraud detection and prevention approach in this context. This study focused on implementing a machine-learning system to identify fraudulent electronic payments. To achieve this objective, an AdaBoost supervised machine learning model was utilised. The effectiveness of the model in accurately detecting and preventing online fraud, thus minimising losses resulting from fraudulent transactions, was evaluated. Different performance measures, including precision, recall, accuracy, F1 score, and latency, were employed and compared with those of other machine learning models, namely CatBoost and XGBoost. A comprehensive assessment of its effectiveness in fraud detection was conducted by comparing the performance metrics of the AdaBoost model to those of other machine learning models. This analysis provided insights into the model's capabilities, strengths, and areas for improvement in accurately identifying and preventing fraudulent e-payments.