The Empirical Research On Credit Evaluation Model Of Peer-to-peer Lending Borrowers Based On Logistic Regression

Since the launch of P2 P network lending,it has developed quickly in China.P2 P network lending has many advantages including low investment threshold,simple and fast operation,and high investment income.The number of online lending trading platforms and online lending transactions have also increased rapidly.However,with the negative events such as the collapse of P2 P platforms,the default of borrowers,and investors having difficulty defending their rights,the confidence of investors has been greatly hit.The number of P2 P online lending platforms and the turnover of online loans have also begun to decline.Among the causes of this phenomenon,the credit default of borrowers is the most prominent.The credit default of the borrower will not only affect the normal operation of the P2 P online lending platform,but also hinder the sustainable development of the entire online lending industry.However,the current domestic scholars' research on the credit risk of P2 P online lending platform borrowers is not perfect.The research mostly analyzes the reasons for the credit risk of borrowers,how to strengthen supervision and other qualitative analysis,but lacks quantitative analysis of credit risk of borrowers.In this paper,a Logistic regression method is used to quantitatively study the influencing factors of the borrower's credit risk,which improves the credibility and interpretability of the model.At the same time,this paper uses information gain for index selection,which excludes indicators that have little effect on the explanatory variables,and improves the calculation efficiency of the model.In addition,this article selects Renrendai,a well-known online loan platform in China,as the research object.Through drawing on the experience of operating the personal credit risk assessment system of traditional commercial banks in China,combining domestic and foreign scholars' research on the personal credit risk assessment index system,and referring to the actual situation of Renrendai,this paper has established an indicator system applicable to China's P2 P online lending platform.The empirical results show that the six factors including borrower's gender,including working time?loan term?average monthly income?accumulated overdue number and credit score.After substituting the test data into the model,it is concluded that the predictive accuracy rate of the borrower's credit risk assessment model for good customers(Y=0)is 97.2%,and the forecast accuracy rate for bad customers(Y=1)is 92.6%.The total accuracy rate is 95.9%.It can be considered that the indicator system constructed in this paper and the constructed credit risk assessment model of the P2 P network lending platform borrower have certain practicability and superiority of prediction effect.Finally,this paper puts forward suggestions for the P2 P network lending platforms: find its own positioning and promote the integration and cooperation between P2 P platforms and the traditional banking industry;improve the information disclosure mechanism of the P2 P network lending platform;improve its own data analysis capabilities and gradually establish a credit system in the P2 P industry;Strengthen employee training and improve employee business capabilities;Also makes recommendations to government regulators: set up an Internet Finance Industry Regulatory Commission;establish a P2 P online lending industry database;strengthen industry supervision and build a hierarchical supervision system.