A Research On Credit Rating Of P2P Lending Platforms In China Based On K-Means Clustering Algorithm Of Factor Score

In recent years,with the rapid development of China's economy,the concept of Internet finance,a new generation,has become increasingly popular.In this context,P2 P network loan,one of the main modes of Internet finance,has effectively met the financing needs of individual customers and small and micro enterprises with its advantages of low threshold,high yield,convenience and efficiency,showing a trend of vigorous development.In 2007,paipaidai,China's first P2 P lending platform,was formally established.Since then,a large number of domestic P2 P loan platforms have been established and entered a period of rapid development,in which the emergence of pleasant loan,renren loan,micro loan network and other well-known P2 P online loan brands.According to the 2017 big data report of P2 P industry released by the first online loan,by the end of 2017,the number of P2 P lending platforms in China has reached 4,065,ranking the first in the world in terms of the number of platforms.However,with the rapid development of P2 P lending platforms in China,risks have also begun to accumulate.Many large P2 P online loan platforms,such as ezubao,shanlin finance and tang xiaofeng,have been hit by lightning cases,which have sounded the alarm bell of P2 P online loan investment to the public.Problems such as delayed payment,difficulty in withdrawing cash,and missing connection in China's P2 P lending platforms have seriously disrupted the order of China's market economy,causing many investors to suffer heavy losses and causing extremely bad social impact.In order to deal with the lack of credit in China's P2 P online loan industry and reveal the credit risks of various P2 P online loan platforms,it is of great significance to carry out independent third-party credit rating for China's P2 P lending platforms.Firstly,based on P2 P network platform loan and credit rating literature at home and abroad on the basis of system analysis,studies the implication and the development of the P2 P network platform loan status,risk characteristics and main problems,at the same time,also expounds the meaning and development course of credit rating and the general method,and points out that the P2 P lending platform credit rating three theoretical basis,which respectively information asymmetry theory,transaction cost theory and game theory.Secondly,this paper studies the credit rating index system and credit rating methods of P2 P lending platforms.In this paper,the design principles of credit rating indicators for P2 P online lending platforms are first formulated,and the credit rating indicator system of P2 P online lending platforms,which includes 2 qualitative indicators and 12 quantitative indicators,is determined based on the existing research on the rating indicators of P2 P lending platforms.The two qualitative indicators are platform background and platform guarantee mode.The 12 quantitative indicators are respectively platform transaction volume,average expected rate of return,average borrowing term,full tender period,outstanding balance,net capital inflow,operating time,proportion of outstanding amount of top 10 borrowers,per capita borrowing amount,number of investors,number of borrowers and capital leverage.After defining the credit rating index system of P2 P lending platform,this paper introduces the K-Means clustering algorithm based on factor score,and takes silhouette coefficient as the evaluation index of clustering quality.Finally,this paper makes an empirical analysis of the research on the credit rating of P2 P network loans in China based on the k-means clustering algorithm based on factor score.In the first step,130 mainstream P2 P lending platforms with a history of more than 18 months were selected,and their monthly data in May 2018 were selected to reduce the dimension of 14 rating indexes of 130 P2 P lending platforms.Four factors reflecting the credit risk of China's P2 P online lending platforms were obtained and scores were calculated.In the second step,K-Means clustering algorithm is used to cluster factor scores of all P2 P online loan platforms for hundreds of times.The silhouette coefficient,the evaluation index of clustering quality,is used to compare the quality of each clustering operation,so as to select the optimal clustering results.Third step,according to clustering algorithm only the classification results of each platform are given to each type of platform's credit rating is given the characteristics of high and low order,based on the mean of each rating index,for all kinds of don't paper summarizes the characteristics of the P2 P network credit platform,determine the types of platform is relatively high and low credit rating,get the end result of a P2 P network platform loan credit rating in China;The fourth step,in order to verify the validity of the P2 P lending platform loan credit rating,this paper deals with the 130 platform for nearly 6 months follow-up,the statistics of each platform is a problem and the cause of the problem,find reduced credit rating as the platform,platform proportion increase in turn,the reality is consistent with the credit rating results presented in this paper.At the same time,this chapter also compares the rating results with those of P2 P online loan rating platforms provided by two well-known P2 P online loan rating websites in China--wangdaizhijia and wangdaitianyan in May 2018,so as to illustrate the advantages of the rating method adopted in this paper.In addition,this paper outs in May 2018 ratings have problems in nine platform,with the same credit rating method,uses data in June 2018 for the rest of the 121 P2 P network platform loan in our country the credit risk rating and inspection,found that reality is still consistent with credit rating results presented in this paper,in this paper,the accuracy of the ratings are still better than the above two rating services to the rating result,further proves the rating method by the use of robustness.There are three innovations in this paper.First,this paper establishes the P2 P lending platform credit rating index system combining quantitative index and qualitative index(platform background and guarantee mode)to enrich and improve the research on the credit rating index of China's P2 P online loan platform.Secondly,this paper uses Python language to implement the K-Means clustering algorithm.It has conducted hundreds of clustering operations on the sample platform data,calculated the silhouette coefficient index and used it as the basis for evaluating the clustering quality,so as to make the clustering results in this paper more accurate.Third,this paper found four factors that reflect the credit risk of P2 P lending platforms in China(operation scale factor of P2 P lending platforms,fund dispersion factor of P2 P lending platforms,security guarantee factor of P2 P lending platforms and profitability factor of P2 P lending platforms),which enriched and improved the research on credit risk of P2 P lending platforms in China.This paper compares the rating results of P2 P online loan with those provided by the two rating websites,wangdaizhijia and wangdaitianyan,to prove the superiority of the credit rating method adopted in this paper.