Research On P2P Online Lending Personal Credit Risk Assessment Model Based On Support Vector Machine

With the increasing importance of credit in actual life,especially the development of Internet Finance(ITFIN),Chinese have quickly entered the era of credit consumption.There are many disadvantages in the process of viewing the borrwer's loan application,such as a long review period,complicated procedures,and high threshold.However,the P2P(Peer-to-Peer)lending has developed rapidly because of its advantages such as a short review period,simplified procedures,and low threshold.However,in order to control risks,it is necessary to make a reasonable assessment of personal credit risk faced with the diversity of borrowers.In recent years,artificial intelligence has been applied to all walks of life,and methods of personal credit risk assessment have been constantly developing and perfecting,ranging from the initial artificial experience to statistical approach,to artificial intelligence.The advantages of Support Vector Machine(SVM)in processing small samples and nonlinear data make it widely applied in the field of personal credit risk assessment.This paper is about how to review personal credit risk based on SVM,which includes the establishment of personal credit risk indicator system,and the optimization of SVM model.In order to construct a more accurate individual credit risk assessment model.we need to find out the right quantity of indicators,because too many or too few indicators are not conducive to the model construction.Model with too many indicators will lead to variable redundancy,which will increase the running time,and even affect the accuracy of the model classification.Model with too few indicators will lack enough variables to classify,resulting in lower accuracy of classification.To solve this problem,the indicators are filtered by means of stepwise regression of independent variables.The selected indicators are used as input ones for SVM modeling.It can be seen from the reference literatures that penalty factor and the kernel function parameters are crucial to the model.From this perspective,some intelligent optimization algorithms are combined to optimize SVM.In this paper,Genetic Algorithm(GA),Ant Colony Optimization(ACO),Particle Swarm Optimization Algorithm(PSO)and Fruit Fly Optimization algorithm(FOA)are used to optimize the parameters of the SVM model,and four personal credit risk assessment models based on SVM are established.The experiments in this paper are based on real data of online loan,and the results show that the proposed method in this paper has a better performance on personal credit risk assessment,which can improve the accuracy of model classification and model operation efficiency.