Research On Risk Identification Of Default Rate Of Car Loan Business In P2P Network Lending Platform

P2P network lending can connect borrowers and lenders directly to conduct online peer-to-peer funds docking.This type of lending model has the advantages of low cost and convenience of online transactions.Therefore it is very popular in the lending market.However,with the sudden increase in the number of P2 P online lending platforms,many problem platforms have begun to massive emerge.Therefore,the regulatory authorities issued corresponding regulatory rules to prompt P2 P lending platforms to shift to the assets with small,decentralized and risk-controllable loan funds,the car loan business become the focus of competition in many peer-to-peer loan platform because of these characteristics.Because business of car mortgage liquidity stronger,the owner of the car can easily turn the vehicles into funds at any time,which makes car become a convenient financial asset.However,there are also many problems such as the moral hazard of borrowers,the risks of mortgaged vehicles and the non-standard operation of platforms.In order to make the car loan business healthy and sustained development,the car loan platform should be well-managed.This paper starts with the risk identification of pre-lending management,uses multiple algorithms in machine learning to build models to study the default rate of the vehicle loan-relief business,and compares the accuracy rate calculated by each model to find out the most suitable vehicle loan.The risk identification model of business default rate helps the P2 P car loan industry to establish a more complete risk identification system.The research on the risk identification of the P2 P vehicle loan-losing business as a default rate can help the car loan platform identify the potential risky borrowers in advance and protect the car loan platform and investors' interests.The focus of this study is to construct a risk identification model for the default rate of vehicle loan-to-deal business.It is select vehicle loan-repayment project data from the weidai platform.First of all,it introduces the process and mode of car loan-receiving business and analyzes it.Then it introduces the car loan-paying platform's operation mode of car loan-relief business,and analyzes the risks of car loan-relief business on these foundations.In the empirical study of building risk model,decision tree,random forest,extreme random tree,GBDT+LR,and XGBoost are used to construct risk identification model.According to the results obtained by each model,the random forest effect is the best among the three tree-based models.The effect of the widely used GBDT+LR model in the credit field is only better than that of the decision tree,and the effect of the XGBoost model is the best.According to the XGBoost research on the factors affecting the default rate of car loan projects,the impact of the mortgage valuation,the number of historical repayment times,mortgage vehicle's mileage and total amount of projects is the greatest.Followed by the number of repayment periods expected,the age,the interest rates,the purpose of the loan,the duration of the project,the purchase price,and the degree of influence of the job type,marital status,sex and income status are very small.This paper uses machine learning algorithm to deal with the data of multidimensional variables,which has a great advantage over traditional P2 P riskcontrol methods.According to the research results,P2 P car loan platform should pay more attention to data resources and strengthen the cooperation with credit agencies.Using big data technology to identify risks can avoid the impact of individual subjective factors,which is conducive to the sound development of car loan business.