An Empirical Study On Enterprise Credit Risk Assessment Of A Commercial Bank Based On Random Forest

In recent years,NPL ratio of Chinese commercial banks continued to increase,risk management situation more and more serious.At the same time,with the big data and data mining technology continues to develop,commercial banks have accumulated a lot of corporate data,the credit valuation technology and dimension are developing continuously,the results are encouraging.This paper not only studies the impact of financial indicators on corporate credit risk,but also analyzes the transaction information,Internet information and commercial loan information that are incorporated into the enterprise.At the same time,this paper briefly introduces the basic principle of a combination model——Random Forest model with stability and strong capacity of noise tolerance,and its advantages compared with other algorithms,and introduces the model to do the empirical research of enterprise credit risk.In this paper,when the empirical research is carried out,the theory and practice are combined to fully consider the practical application of the model in the risk management of commercial banks,rather than the simple theoretical analysis.In this actual situation,the paper constructs a new composite index(coverage – accuracy ratio)to adjust the model and compare it with other models.The experimental situation shows that the random forest model combines the accuracy and the coverage ability,which is more suitable for the business risk management of commercial banks.It is hoped that through the research and practice of this paper,it will provide some reference for the risk management of Chinese commercial banks.