Risk Influencing Factors And Prediction Analysis Of P2P Network Loan Platform

Over the years,the combination of Internet technology and finance has led to the emergence of Internet finance,which reduces transaction time and financing costs,making the transaction process simpler and more convenient.Especially,the emergence of P2 P network lending platform alleviated the difficulty and expensiveness of personal loans and financing of small and micro enterprises.P2 P network lending platform also made up for the credit gap of traditional credit.However,the risks of online lending platform are increasing with the rapid development of the P2 P online lending platform industry,due to the uneven quality of the platform itself and the lack of risk control capabilities,unbalanced regional development,the lack of perfect supervision and management programs of the state and relevant regulatory authorities.The platform problems emerge endlessly,and the platform bankruptcy trend also erupts,which greatly threatens the financial security of investors and thus makes the whole industry face a huge credit crisis and risk.Through Python Network crawler technology,this dissertation crawls the relevant data on the platform network and constructs the index system.Through Kaplan-Meier curve and logrank test,the survival probability of sample platforms corresponding to different groupings of each classification variable is found to be significantly different.Lasso variable selection method is used to screen out variables that have significant impacts on the survival status of P2 P platform.Moreover,Cox model is established to quantitatively give the impact direction and degree of each variable on the risk of P2 P platform problems.The results show that protective factors including: In addition to the comprehensive yield variables,the platform background,the number of target types,investment duration,bond transfer,whether the target period is more than 12 months or more,regulatory associations,bank deposits,safeguards,home visits,physical store photos,cash withdrawal scores,and the proportion of recommended people are all protective factors.The larger these indicators,the less likely the P2 P platform is to have problems.Furthermore,the platform is predicted and classified.Finally,the survival tree model is established to visually observe the classification basis of the model variables.Compared with the classification results of Cox proportional risk model,the survival tree prediction model is better than Cox prediction model.Introducing the survival analysis model into the risk study of P2 P online lending platform can accurately quantify the impact of various factors on the platform risk,dynamically show the process of the survival rate of the problematic platform declining continuously with time,and then make a classification forecast for the platform.Through the empirical analysis,this dissertation aims to provide investors with a method of identifying the risk degree of P2 P online lending platform to reduce the investment risk.Additionally,the dissertation provides platform operators with an early warning means to identify and prevent the problems of P2 P online lending platform to improve the platform's ability to resist risks.Finally,with all the above analysis,regulators may formulate regulatory policies to provide reference and standardize the online lending industry,so that prevent and control the risk of the online lending industry more effectively.