Prediksi Non Performing Loan (NPL) pada Debitur Kredit Pemilikan Rumah (KPR) menggunakan XGBoost

Abstract

Bank Sumut mortgage debtors often face fluctuations in collectibility status influenced by various economic and personal factors. To overcome this problem, the XGBoost method is used because of its effective ability to handle classification problems and provide accurate predictions. This study aims to develop a prediction system that can help Bank Sumut in improving the efficiency of credit decision making and risk management. In this study, the XGBoost model was built and trained with data covering factors such as ceiling, ending balance, interest, principal, and product name. The results showed that this model successfully predicted the debtor's collectibility status with a very good accuracy rate of 97.31% using 80% training data and 20% test data. With an accurate prediction system, Bank Sumut can be more effective in managing its credit portfolio and making more appropriate credit decisions. This research is expected to contribute to the advancement of science and technology in the banking sector, especially in credit analysis and risk management.