| With the rapid development of various Internet platforms such as e-commerce,social networking,Internet finance,and P2P consumer credit and consumer finance,the short-term board of the central bank is increasingly prominent in its timeliness,comprehensiveness,and hierarchy.How to dig deeply into the massive information flow of the Internet,develop a risk control model for big data based on massive indicators,comprehensively assess the credit risk status of corporate customers,and provide a basis for judging the financial credit approval of P2P lending platforms,which has become the current new generation credit risk model system.The core topics of construction.The research in this article is a useful attempt to consider the senior management information of the company and provide financing for SMEs in this context.On the basis of theoretical research and literature review,this paper carries out risk identification for the SME credit rating system of P2P financing lending business.Based on the credit evaluation indicators of traditional commercial banks,a portrait of a P2P corporate client user under consideration of corporate executive information was established.The main conclusions are as follows:(1)Constructing a P2P corporate customer credit risk evaluation index system considering corporate executives'informationFirstly,this paper constructs the P2P personal credit evaluation system from six aspects such as identity verification information.On this basis,the personal credit information of senior executives of enterprises was taken as a supplement to the corporate credit information,and together with the company's own situation and external evaluation information,the P2P corporate customer credit risk pre-selection evaluation index system was jointly constructed.Then,through the evaluation of WOE code and value evaluation indicators,the final establishment of P2P corporate customers'credit risk evaluation index system.(2)Constructed a P2P corporate customer credit risk assessment modelTaking into account the massive information flow of the Internet,the feature extraction of indicators is particularly important in the credit assessment model.In this paper,the horizontal dimension reduction method(PCA)and nonlinear dimension reduction method(RBM)are compared horizontally.Then the SVM,Logistic and KNN models are selected in the traditional machine learning classifier.The empirical results show that the RBM effect is better than that of the PCA when the same model is selected.Under the same credit risk assessment index system,the Logistic model gives the highest accuracy of the SME credit status classification;this objectively illustrates Under the evaluation index system constructed in this paper,the P2P network lending platform using the RBM-Logistic model for the credit risk assessment of P2P corporate customers is scientific and reasonable.At the same time,taking into account the difference in losses caused by misclassification,the probability of sampling the“wrong customer with poor credit rating as a good customer”is revised to be (?).A penalty factor was introduced to correct the combined weights of the combined classifiers,an improved G-AdaBoost algorithm was implemented,and the classification accuracy of the final model was improved.(3)Propose P2P network platform to supervise and advise corporate customers'credit riskCombined with the theoretical research and empirical research results of this paper,this paper proposes a P2P network lending platform to evaluate the credit risk of SMEs from three aspects:credit risk assessment work,credit risk assessment support work,and protection against the legal risks of P2P network lending platform. |
