Credict Risk Assessment Of P2P Borrowers Based On Multiple Classification Models

In recent years,P2P lending has been developing rapidly as an important model of Internet finance.P2P lending overcomes the shortcomings of bank lending and private lending.It can make full use of the idle funds of the society,promoting the marketization of interest rate,and improving the efficiency of social resource allocation.P2P lending has become more professional than before,we should pay more attention to the safety of lending and the risk control of lending behavior.This paper studies the credit risk of P2P borrowers based on multiple classification models.Firstly,it was introduced the theory of each classification model.Then,data preprocessing was performed to prepare for subsequent modeling.Then,the logistic regression,random forest,GBDT and combinatorial model were used for risk assessment.Finally,the models were evaluated and summarized.Based on the data of two years of Lending Club,this paper finds that the best model is the combinatorial model.At the same time,this paper proposes to promote the standardization of domestic credit information system,providing reliable data for individual credit evaluation.In addition,the study can be used as a reference method for the credit risk assessment of P2P lending platforms in China.