Assessment Of Credit Risk Of P2P Network Loan Borrowers

With the advent of the era of big data,the Internet finance industry has gradually developed.Under the current environment of inclusive finance,Internet finance and traditional finance form a new financial model.Third-party payment platform,P2 P online loan platform and crowdfunding equity platform are the main Internet financial models in China,especially the P2 P online loan platform has developed rapidly in China in recent years.P2 P online lending platforms in China have a great demand for borrowers’ credit information,but currently there is no compliant source to obtain the borrowers’ credit information,which makes online lending platforms bear great operational risks.Network platform in for all kinds of different credit loan borrowers to lend,investors have different decision-making mistakes of acceptance is different,which if the credit rating low customer misinterpret the credit rank higher,the principal and interest of the investors will face a big loss,but if the credit rating higher customer mistakenly judgment for lower credit rating of customers,to guarantee the investors’ principal,though,but at the same time lose part of the interest rate.In view of the above realities,with all the personal loan lending data in this paper,the empirical part as an example,the basic characteristics of the P2 P network credit operation mode,fully balancing the interests of investors and investment risk,the value of different categories of borrowers to select the appropriate credit risk assessment model,in order to reduce the credit risk of the borrowers,maximize the safety net credit platform,and the interests of investors.The research of this paper is mainly carried out from the following aspects:First according to the borrower of loan,loan number,loan amount between clustering analysis based on RFM model,the borrower can be divided into important to keep customers,develop customers,clients and generally low value customer,fully explore mining borrowers distribution characteristics and change law of intrinsic value,effectively improve the precision and efficiency of P2 P platform in the borrower’s management.Secondly the collected 15 related to borrower’s credit rating index quantitative and standardized processing,due to the importance of characteristic value division precision directly affect the borrower credit risk assessment,so this article with random forests on relatively important degree of characteristic variables,combined with the correlation test,used to build credit indexes evaluation system has nine characteristics of variables.Then appropriate risk assessment models are selected for the four types of borrowers with different value categories.Random forest model and support vector machine model were selected as risk assessment models,and the accuracy,precision and recall rate of the model were used as the evaluation criteria.The research results show that random forest model is more suitable for evaluating borrowers and important development of the credit risk of the borrowers,can reduce the low credit rating of the borrower as high credit rating classification error,the interests of the investors’ losses can be small investment and avoid most of the investment risk,thus maximizing the interests;The support vector machine(SVM)model is more suitable for evaluating the credit risk of ordinary borrowers and low-value borrowers,which can reduce the misclassification of borrowers with high credit rating into those with low credit rating,so that investors can realize a small part of returns while avoiding risks.Finally,according to the four types of borrowers,the comprehensive prediction accuracy of the whole borrower is calculated,and compared with the single model,the prediction accuracy of all borrowers is improved to a certain extent.Based on the actual operation mode of P2 P lending platforms in China and the assessment results of borrowers’ credit risks,Suggestions are proposed from four perspectives,namely,regulatory authorities,P2 P lending platforms,investors and borrowers.