Comparative Study On Credit Risk Of P2P Online Loan Borrowers Based On Several Common Models

In recent years,with the advantages of convenient operation and low investment threshold,the number of online lending platforms and the scale of transactions have increased year by year.In the current context of Internet finance,it has become a representative investment model.The rapid popularization and development of China's P2P online lending industry has played an active role in facilitating residents'borrowing,improving the development of small and micro enterprises and enriching China's multi-level financial markets.Of course,while facing opportunities,there are also many risks,such as credit risk,technical risk and legal risk,among which the most critical one is credit risk.In the P2P industry,credit risk is the borrower's default risk.In order to ensure the improvement of core competitiveness of various online lending platforms,P2P online lending is moving towards a mature direction of development,and it is urgent to solve the problem of credit risk control.In view of the background and current situation of the development of P2P online lending in China,this paper measures the borrower's credit risk based on the strong market supervision environment and the absence of loopholes in the legal corner.The main research work is as follows:Firstly,in order to construct the borrower's credit risk assessment model reasonably and effectively,this paper establishes a model based on the traditional credit risk assessment index and the platform's own characteristics,aiming at the imperfect credit reporting system in China.It mainly includes borrower's basic information,work information,credit information,asset information and borrowing information,and constructs a new feature repayment ratio based on the original indicators to verify that this variable is of great significance to model fitting.Secondly,combining the current situation and characteristics of P2P network lending platform in China,this paper analyses the credit risk and causes of borrowers,and constructs a new evaluation portfolio model.The credit evaluation model based on this model not only has the advantages of Stochastic Forest equilibrium error,but also has the characteristics of regression to solve multiple collinearity problems.Thirdly,comparing with the common risk control models.Using the real data of Renren Credit and the results of various evaluation indicators,it can be shown that the model-based credit risk assessment model of borrowers has higher classification accuracy than regression,random forest,kNN algorithm and so on.Finally,through the analysis of characteristic contribution degree of regression and random forest,it is found that "credit information","loan information" and "professional information" are the key indicators to evaluate the credit risk of borrowers,which can optimize the credit risk management framework.