Article Per Year (5 Year)
p-Index From 2020 - 2025
0.23
P-Index
This Author published in this journals
All Journal International Journal of Economics Development Research (IJEDR)
Andri Ismatullah Gani
Universitas Diponegoro
Economics, Econometrics & Finance Social Sciences

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

 1 
Credit Default Prediction Model Using Machine Learning for Credit Monitoring: Empirical Study on Banking in Indonesia Andri Ismatullah Gani; Irene Rini Demi Pangestuti
International Journal of Economics Development Research (IJEDR) Vol. 5 No. 3 (2024): International Journal of Economics Development Research (IJEDR)
Publisher : Yayasan Riset dan Pengembangan Intelektual

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37385/ijedr.v5i3.4214

Abstract

Credit default is the failure of a borrower to make required principal or interest repayments on a debt. In credit risk management, it is important for banks to anticipate credit defaults, whether in the credit underwriting process or in the area of credit monitoring. In this study we focus on the use of machine learning in the area of credit monitoring to predict default of working capital credit and investment credit based on non-demographic debtors’ data, where we then test model’s accuracy and level of precision that can be achieved, and identify variables that have high importance on the predictive model. Finally we evaluate the model in terms of their explainability using eXplainable Artificial Intelligence (XAI) tools to identify the relationship of the variables with credit default. We use monthly snapshot of sampled credit accounts data from 105 banks in Indonesia for the period of August 2018 to December 2019 to build a classification machine learning model, and finally evaluate the model in terms of their explainability using SHapley Additive exPlanations (SHAP) as one of the algorithms for eXplainable Artificial Intelligence (XAI) tool. The machine learning model we built can achieve 98.85% accuracy overall and 75% precision in predicting true positive or correctly predicting credit is defaulted. And by using SHAP, we can understand how each variables contribute to the model’s prediction result and thus its relationship with credit default, where most are consistent with previous researches.
Co-Authors Irene Rini Demi Pangestuti