Borrowers’ credit quality scoring model and applications, with default discriminant analysis based on the extreme learning machine

Abstract

Through evaluating the weight of evidence method and calculating the information value (IV), this article proposes a method to evaluate the credit qualities of borrowers based on the extreme learning machine, the fuzzy c-means (FCM) algorithm, and the calculation of a confusion matrix. Through screening credit rating indexes, we established a credit scoring model of the borrower. In addition, we constructed formulas to determine the probability of default and default loss rate. The model also classifies the credit qualities of borrowers. In addition, we designed a selection algorithm for the borrower's credit quality rating index, and a borrower's credit quality rating algorithm. This paper collects sample data of 7706 borrowers of Renren loans from the Internet. The credit scores of the borrower, the default probability, and the default loss rate of each type of borrower are calculated, and the repayment status of borrowers are analyzed. We divided the borrowers into 7 grades and 5 grades by calculating a confusion matrix. The experimental results show that the overall accuracy of the credit scoring model is 98.5%, in which the accuracy for non-default samples is 98.9%, and the accuracy for default samples is 88.3%. The accuracy of the established credit quality rating model proved to be relatively high, and it can provide important reference values and scientific guidance for banks, financial institutions, and major financial platforms. It can also judge and predict default behavior.