Credit Scoring for Financial Services Institution using Ant Colony Optimization Algorithm under Logistic Regression Model

Abstract

Economic and trade activities are important in a country. All these activities are regulated by financial institutions, such as banks. The process of channeling funds to the public or known as credit is one of the tasks of the banking sector which aims to improve the people's economy. Credit granting is required for credit analysis, which is useful to determine the level of eligibility of a debtor to receive credit. The function of the credit analysis is to reduce the credit risk of prospective debtors who have failed to pay as well as to avoid financial institution losses or charges. The method used to analyze credit risk in this study is the Ant Colony Optimization algorithm in the Logistic Regression model. Past data held by each prospective debtor obtained from one financial institution in Indonesia is used as a feasibility parameter in this analysis. The results of the study showed that eight variables analyzed were five variables including the significant influence (age of debt ( X1 ), family dependents ( X 2 ), value of the collection ( X 4 ), the number of credit limits ( X 6 ), and the term of the loan ( X8 ) while the other three variables (the amount of savings ( X 3 ), income per month ( X 5 ), net income ( X 7 ) are not significant to the risk of default.