High-dimensional Data-driven Credit Risk Evaluation Of Online Loan

With the advancement of Internet technology,financial instruments continue to innovate,and people's consumption concepts are constantly changing.The traditional financial lending model has been unable to meet people's daily needs.In order to meet the needs of social progress,economic development and people's lives,with the rapid development of the Internet,online lending has emerged.In recent years,the classification of online loan defaults has been a hot issue for scholars.On the one hand,the increase in the amount of data allows researchers to have more reference in assessing the credit risk of online loans.On the other hand,with the advancement of big data technology,the dimension of the influencing factors of credit data sets is becoming higher and more complex.The online loan data presents a complex feature of high dimensionality,which makes some original credit evaluation methods not applicable.The high-dimensional loan default data set puts forward higher demands on the flexibility of the evaluation method.For this purpose,this paper has carried out research work on the high-dimensional characteristics of online loan credit data sets.The main research contents are summarized as follows.First of all,a comprehensive credit evaluation index system for online loan people is constructed.This paper comprehensively considers the credit evaluation index system of representative online loan platforms in China and abroad,as well as the scholars' research on personal credit evaluation.Based on the characteristics of high-dimensionality of online loan credit data sets,the current online credit evaluation indicators were screened.Considering that online loan customers are mostly young people,this paper checks the three screened indicators,including basic personal information,asset information,and working information.At the same time,from the perspective of big data relevance,this paper fully considers the relationship between social platform behavior and personal credit.The“social platform information”indicator was added to ensure the applicability of the online loan credit evaluation index system.Secondly,in terms of the high-dimensional data-driven idea and related research,this paper proposes and constructs a multi-order combinatorial dimensionality reduction method.Subsequently,this method is used to conducts an empirical analysis on online credit evaluation.The method is divided into three main steps.In the first step,the most commonly used linear and nonlinear feature extraction methods are selected as the basic methods of multi-order combinatorial dimensionality reduction method.In the second step,in terms of the advantages of linear and nonlinear dimensionality reduction methods,different single dimensionality reduction methods are combined into constructing different multi-order combinatorial dimensionality reduction methods.In the final step,through the optimal multi-order combinatorial dimensionality reduction method,the online loan credit is classified.The empirical results show that the multi-order combinatorial dimensionality reduction method constructed in this paper can effectively reduce the dimension of high-dimensional credit data sets.According to the data after dimensionality reduction,the credit classification results are better and more reliable.Finally,this paper constructs a combinatorial dimensionality reduction method based on feature selection and feature extraction,and uses this method to study the credit evaluation of online loans.This paper compares the difference between feature selection and feature extraction in data dimension reduction and summarizes the advantages and disadvantages of the two methods.Then,a combinatorial dimensionality reduction method is proposed,and the classification results of online loan default are compared and analyzed.The method is also divided into three main steps.In the first step,three feature selection methods and two feature extraction methods are selected as the basic methods.In the second step,in terms of the advantages of feature selection and feature extraction,with the aim of improving the classification results,a combinatorial dimensionality reduction method is constructed.In the final step,through the optimal combinatorial dimensionality reduction method,the online loan credits are evaluated and classified.The empirical results show that this method can obtain better classification results of online credit.In summary,this paper investigates the credit risk problem of online loans,including the construction of a credit evaluation index system for online loan people and the construction of two combinatorial dimensionality reduction methods for online loan credit classification.The empirical analysis shows that the methods constructed in this paper can obtain better dimensionality reduction results and can optimize the credit risk classification results.