Research On Credit Risk Assessment Of Commercial Banks

In recent years,credit risk has been continuously released,and the rate of non-performing loans of commercial banks has increased rapidly.Under such circumstances,how to effectively manage credit risk by using cutting-edge computing technology in commercial banks is an urgent research topic.Based on the customer default data and the actual situation of China's commercial banks,the logistic regression model,the random forest model and the decision tree model are established to evaluate the credit risk accurately.From the prediction recall rate,the recall rate of decision tree and logistic regression for predicting bad customers is 0.38 and 0.42 respectively,while the recall rate of random forest model is only 0.3,which indicates that random forest model is not easy to grasp potential bad customers.In terms of model prediction accuracy,the accuracy rate of decision trees and logistic regression to bad customers is only 0.14 and 0.07,while the prediction accuracy of random forest to bad customers is up to 0.63,which indicates that most of the bad customers predicted by the random forest model are correct.From the K-S value,the K-S value of the random forest model is the highest,which is0.3216,which indicates that the random forest model is more capable of distinguishing good and bad customers.After comparing the prediction results of a single model,the three models are also trained in undersampling,and they are used as subclassifiers to integrate the new prediction sequences according to the prediction results of each sub classifier.It is found that the prediction effect of ensemble prediction model is better than that of single model,which provides a new idea for credit risk assessment.The empirical results obtained in this paper have strong operability and applicability,and are easy to popularize.The risk control department of commercial banks can make use of the customer default data of the bank after the loan to build a credit risk assessment model suitable for your own bank according to the train of thought in this paper.