Online Payment Fraud Detection Optimization with XG Boost and Recursive Feature Elimination
Jati Sasongko Wibowo, Budi Hartono, Veronica Lusiana · Journal of Software Engineering and Simulation · 2024
Online payment fraud is an increasingly pressing issue as the volume of digital transactions grows. Accurate and fast detection is essential to minimize financial losses. This paper presents an approach to optimize fraud detection using the XGBoost algorithm and the Recursive Feature Elimination (RFE) feature selection technique. In this research, we use an online payment fraud dataset to train a model that can distinguish between legitimate and fraudulent transactions. The main contribution of this research is to demonstrate the effectiveness of the combination of XGBoost and RFE in improving fraud detection performance. The methods used include data preprocessing, feature selection with RFE, and model training with XGBoost. The evaluation results showed that the XGBoost model with RFE achieved a precision of 0.96, recall of 0.86, and f1-score of 0.91 for detecting fraudulent transactions, with an overall accuracy of 99.98%. In conclusion, the use of XGBoost together with RFE feature selection proved to be an efficient and effective approach for fraud detection in online payment systems, providing a reliable solution for real-world applications in the financial industry.