Research On Running Forecasting Model Of Peer To Peer Online Loan Platform With Linguistic Analysis

Since the 18 th National Congress,the importance of preventing and controlling financial risks has been repeatedly emphasized.Keywords such as risk prevention and vigorous development of inclusive finance have also frequently appeared in the highest-level conferences and occasions in China.At the Fifth National Financial Work Conference on July 14,2017,General Secretary Xi Jinping once again emphasized the importance of risk prevention and financial supervision,and proposed to accelerate the construction of an inclusive financial system.The online loan industry,as a high-risk area for financial risks and an inclusive financial carrier,since 2016,platform risk incidents have frequently erupted,and platform running events are not uncommon;In 2017,under the pressure of the most stringent regulatory year in history,a large number of non-compliant online lending platforms have been rolled out,or they have fled,but some of the platforms have turned from dark to dark,becoming an invisible risk platform that continues to attract investors to invest.At present,China's online lending industry supervision is in an adjustment period,various regulatory policies are in the experimental stage,and the online lending industry is still unclear.At this stage,how can investors avoid the “minefield” and choose a high-quality platform for investment;how P2 P platforms to optimize its service and highlight its own advantages;how the regulatory authorities find the final balance point in the chaos and improve the supervision mechanism of the online loan industry has become an important concern at present.At present,China's online lending industry supervision is in an adjustment period,various regulatory policies are in the experimental stage,and the online lending industry is still unclear.At this stage,how can investors avoid the “minefield” and choose a high-quality platform for investment;how the P2 P platform to optimize its service and highlight its own advantages;how the regulatory authorities find the final balance point in the chaos and improve the supervision mechanism of the online loan industry has become an important concern at present.Due to the chaotic development of the online loan industry in the early stage,it is difficult to obtain transaction information and financial information of all platforms at present,and it is impossible to evaluate and predict the operation of the platform through such hard information;Therefore,this paper takes a different approach,The text mining technology is applied to the construction of the risk prediction model of the online loan industry.The theme model is used to extract the theme and construct the subjective index quantitative text information to compensate for the opaque data quality and uneven quality of the online loan industry.Based on the information characteristics of P2 P platform,this paper proposes a P2 P online lending platform running prediction method incorporating text information.Firstly,This paper uses natural language processing technology to preprocess the evaluation text,and uses the topic model to extract and quantify the relevant variables in the user evaluation text,and then analyzes the relationship between different variables and the platform "running".The results show that the probability of running the platform in other regions is 60% higher than that of the platform in the municipality;The platform background has a significant impact on the platform running,and the private platform and the no-background platform are easier to run than other platforms;The platform capital has a significant impact on the platform's running,but it can't just look at the platform's registered capital.It should pay attention to the platform's paid-in capital.The results show that the probability of running on the platform with less than half of the registered capital of the platform is 26% higher than that of the platform with more than half of the registered capital;The existence of bank depository,accession to the regulatory association,and ICP certification can significantly reduce the probability of platform running;At present,there is a situation in which the demolition of the platform increases the liquidity of the target,which will lead to the mismatch of the funds of the platform and increase the risk of the platform closing the road.The result also confirms this phenomenon;In the risk protection model of platform investors,in addition to the platform advancement,the other guarantee modes help to reduce the possibility of platform running,especially the insurance institution guarantee mode.The online lending platform that does not use insurance institutions guarantees is 294% higher than that of the platform with insurance institutions guarantees.Obviously,the platform with insurance guarantee is conducive to platform development;The user sentiment variables and most of the subject variables obtained from the user evaluation are enough to reflect the platform's running risk.In the positive evaluation,the user may be more concerned about the platform return rate,platform qualification,platform service and fund standing.In the negative evaluation,users are more concerned about cash withdrawals,platform fraud,platform marketing and funding.Secondly this paper designs a two-stage variable selection method to combine the basic information variables and text information variables,and on this basis,compares the accuracy of Logit regression prediction model with machine learning prediction model.The results show that the road prediction model based on machine learning P2 P online lending platform has significantly improved the prediction effect after incorporating text information.