Research On The Risk Of P2P Lending Platform

With the vigorous development of Internet finance,it brings a variety of new financial services.P2 P network lending is one of the most important embodiment of Internet finance,which has been popular in the market.However,the uneven quality of lending platforms and the early loose regulatory environment have brought great challenges to investors' rights and the development of the industry.Due to frequent thunderbolt in P2 P lending platform,This dissertation analyzes the risk of P2 P lending platform from two aspects: qualitative and quantitative,so as to realize the early warning of P2 P platform.On the one hand,in the qualitative analysis of risk early warning of P2 P lending platform,we collected 2725 lending platform data from network,obtained the relevant characteristics of the lending platform's qualification after data preprocessing and other preparatory work,and then combined with natural language processing technology to mine the characteristics from the text data of user evaluation.In the analysis of modeling experiments,The single model and the integrated model stacking are selected for the comparative experiment to qualitatively work on the platform,that is,to predict the lending platform is a problem platform or a normal platform.On the other hand,in the quantitative analysis of the risk of lending platform,this dissertation proposes an improved weighted similarity method,and combined with the risk assessment characteristic matrix established in the experiment to calculate the risk value of each lending platform.The experimental results show that,in the qualitative process of the platform,the addition of "user comments" feature can effectively improve the performance of the model,and in the three natural language processing technologies,the risk early warning model based on Word2 Vec performs better,with an accuracy of 89.93% on817 test samples.in the qualitative analysis of the platform,the single model and the integrated model stacking are also compared in the process of quantifying the risk value of the platform.In the process of quantifying the risk value of the platform,the results show that the risk value of the risk early warning model based on Word2 Vec in the problem lending platform is significantly higher than the normal lending platform,and it can more effectively mine the high-risk lending platform.