The best way to conference proceedings by Francis Academic Press

Web of Proceedings - Francis Academic Press
Web of Proceedings - Francis Academic Press

Research on Credit Risk Assessment of Loan Customers

Download as PDF

DOI: 10.25236/icemeet.2020.044


Ming Jin

Corresponding Author

Ming Jin


With the continuous development and progress of China's market economy, a special economic mode has emerged gradually in the commodity economy, that is, credit economy. Credit has multiple levels and many side meanings. Credit in the economic sense is the activity of borrowing and lending. It is a special form of value movement under the condition of repayment. Under the condition of commodity exchange and currency circulation, The creditor borrows money or credit sales in the form of conditional transfer. The debtor pays the loan or pays the loan at the agreed date, and pays interest. The most important condition in the credit economy is to abide by the credit agreement. Otherwise, it will produce credit risk. In developed countries, risk has become a constituent factor of the business objective. A business item, a loan, often has high income and large risk, low income and low risk. Credit risk refers to the possibility that banks suffer losses because of the uncertainty in credit activities. In fact, it is all risks caused by customers' default. For example, the quality of assets caused by the failure of the borrowers in the asset business to deteriorate, and the depositors' large withdrawals of cash in the debt business and so on. In the field of credit evaluation, more and more mathematical methods and technologies have been used to achieve more precise and interpretable purposes. In this paper, based on the historical information of bank loan customers, the machine learning algorithms are introduced into the modeling of personal credit evaluation, and the data mining decision tree (CHAID algorithm) and neural network are used. According to the characteristics of different customers, we can distinguish the customers who are in good credit condition or those whose credit condition is not good. Then we compare the models obtained and choose an optimal model, so that banks can predict the possibility of future lenders default on loans.


Credit evaluation, Chaid, Artificial neural network, Model evaluation