Classification Model for Online Payment Transaction Fraud Using a Decision Tree Based on the Gini Index

Abstract

The development of digital technology has led to a significant increase in the use of online payment systems. However, this rise in transaction volume has also resulted in heightened risks of fraud, which threaten the security of digital transactions. Early detection of fraudulent activities is essential to prevent financial losses and maintain user trust in these payment systems. This study aims to build a classification model for detecting fraudulent online payment transactions using the Decision Tree algorithm based on the Gini Index. The dataset used consists of digital financial transactions, which include both numeric and categorical features. The model's performance is evaluated using confusion matrix metrics such as accuracy, precision, and recall. The results indicate that employing the Gini Index for feature selection enhances the model's performance, achieving an accuracy of 93.42% and a notable increase in recall for minority classes, such as DEBIT transactions. The Gini Index-based Decision Tree has proven effective for the interpretive and efficient detection of fraudulent transactions. This study contributes to the development of a more accurate digital fraud detection system that can be implemented in real-world online payment systems.