Predictive Modeling for Fraud Detection in Mobile Payment Systems with Decision Tree Method

Abstract

The rapid growth of digital technology has significantly increased the use of mobile payment systems as a fast, practical, and efficient financial transaction method. However, the growing volume and frequency of transactions have also led to a higher risk of fraudulent activities, which may cause financial losses, reduce user trust, and threaten the security of digital payment platforms. Detecting fraudulent transactions is a critical issue because transaction data is typically large, complex, and continuously generated. To address this problem, predictive modeling using machine learning can be applied as an effective solution. This study aims to develop a fraud detection model for mobile payment systems using the Decision Tree method. The research method includes data collection, preprocessing, feature selection, data splitting, model training, and evaluation using performance metrics such as accuracy, precision, recall, and F1-score. The expected results of this study are the successful classification of fraudulent and legitimate transactions with satisfactory performance, as well as the identification of the most influential features in fraud detection. This research is expected to contribute theoretically to the application of machine learning in fraud detection and practically to support financial technology providers in improving transaction security, decision-making processes, and more effective fraud prevention strategies in modern digital payment ecosystems.