Research On Credit Risk Assessment Of P2P Lending Borrower Considering Auditing Information

P2P network lending has gained explosive growth in recent years due to its high returns,low entry barriers,and convenient and fast transaction.It is considered as a representative investment and financing model under the Internet financial environment.The development and popularization of P2 P network lending are conducive to meeting the financial needs that traditional financial institutions cannot cover,and are of great significance for enriching and improving Chinese multi-level financial markets.However,due to the limitations of information asymmetry,low default costs,complexity of creditor's rights and the lack of a good social credit system in China,P2 P network lending has brought in various problems and risks in the development of the industry.Among them,the phenomenon of default by borrowers is an important obstacle to the development of the P2 P lending industry.In view of this situation,this paper studies the credit risk of P2 P lending borrowers,mainly from the following aspects:First of all,this paper proposes a credit risk assessment index system for borrowers that includes basic information,job information,credit information,asset information,borrowing information,and audit information,and then uses the borrowers data of Renrendai platform to verify that audit information can improve borrower's credit risk assessment effect.Secondly,a credit risk assessment model for borrowers based on GBDT-SVM was constructed.Borrower credit data has the characteristics of complex distribution and many features.Considering that gradient boosting decision tree(GBDT)can quickly select distinctive features,support vector machine(SVM)has strong generalization ability.this paper constructs a borrower's credit risk assessment model combined with GBDT and SVM.First,use GBDT to extract effective information from the borrower's original data to construct new feature combinations,and then use the SVM model to perform risk assessment on the borrower based on the new feature combination.This will not only simplify the structure of the SVM,but also improve the classification accuracy of the model.The new model combines the advantages of both GBDT and SVM.Finally,the proposed borrower credit risk assessment method is compared with four common risk assessment models: logistic regression(LR),artificial neural network(ANN),SVM,and clustering algorithm.The empirical results based on borrowers data from Renrendai show that the P2 P lending credit risk assessment model for borrowers based on GBDT and SVM has a higher classification accuracy.Then,the characteristic contribution of the indicators is further analyzed.The research results show that credit information and borrowing information are the key indicators for assessing the borrower's credit risk.