基于Logistic回归模型的P2P借款人信用违约风险评估模型研究

Research on the Credit Risk Assessment of Peer-to-peer Lending Borrower Based on Logistic Regression Model

针对P2P网络信贷平台信用风险特点,以借款人违约情况为被解释变量,运用Logistic回归方法建立借款人信用违约风险评估模型。原始数据从人人贷网站抓取获得,选取的原始评估变量有24个,通过信息增益进行指标降维,得到19个解释变量,并以此建立了Logistic回归模型。通过五步逐次回归得出,性别、逾期次数、逾期金额、身份认证和学历认证等5个因素作为评价个人信用风险的主要依据,并建立了Logistic回归模型。回归模型的判别准确率表明所构建的借款人信用风险评估模型预测效果较好。

Based on the credit risk characteristics of P2 P network credit platform,this paper uses Logistic regression method to establish a credit default risk assessment model for borrowers when taking the borrower default situation as the explanatory variable.Raw data is obtained from the crawl of the Renren loan website.There are 24 original evaluation variables selected.Through the information gain,the explanatory variables are reduced to 19.Through five steps of regression,the five factors including gender,overdue number,overdue amount,identity certification and academic qualifications should be used as the main basis for evaluating personal credit risk,and the logistic regression model is established.The discriminative accuracy of the regression model indicates that the constructed credit risk assessment model of the borrower has a better prediction effect.