Credit scoring model enhancement for personal bankruptcy prediction in Malaysia: towards achieving debt sustainability / Aqilah Nadiah Md. Sahiq

Generally, the idea of getting bankrupt never crosses our minds, but being declared bankrupt can happen to anyone with some form of debt. Several critical issues warrant great attention such as the prevalence of personal bankruptcy incidence, high enrolment in the Debt Management Program (DMP), and high debt service ratio (>30%) among Malaysians. These indicate that many Malaysians are in financial distress, lack financial preparedness, and are in high need of debt repayment assistance. If these issues are left unaddressed, they may lead to many Malaysians going into bankruptcy. These key issues underscore the importance of developing an efficient credit scoring model to address the concern that majority of Malaysians are in financial distress. In this context, the objectives of this study are to identify the key determinants capable of predicting the likelihood of personal bankruptcy in the future, develop personal bankruptcy credit scoring models, and compare the models’ performance. This study focused on microeconomic indicators using 31,200 samples of the DMP dataset for a period between 2016 to 2020. Using RapidMiner software, the methodology for this study consisted of the application of logistic regression (LR) and support vector machine (SVM) models through the adoption of Cross-Industry Standard Process for Data Mining (CRISP-DM) framework.