基于二分类LR的个人信贷违约风险评估

Personal Credit Default Risk Assessment Based on Binary LR

本文采用阿里云网站天池实验室中的公开部分个人信贷面板数据资料,利用STATA软件实现二分类Logistics Regression建模,对个人信贷信用风险进行了研究。研究结果显示:该模型的拟合能力较好,对违约的识别正确率较高,达到80.26%。在0.95的置信区间内,工龄、信用卡负债和负债率对信用违约风险有显著的影响,而其他的因素的影响不是很明显。借款公司可借助该模型评估贷款客户的违约风险,改善贷款的质量。

The published personal credit panel data in tianchi laboratory of aliyun website were used to study the credit risk of personal credit in this paper,and a dichotomy Logistics Regression model was implemented by using STATA software.The results show that the model has a good fitting ability and a high recognition accuracy of default(80.26%).Within the confidence interval of 0.95 years of service,credit card debt and debt ratio had significant effects on credit default risk,while other factors were less significant.Loan companies can use this model to evaluate the default risk of loan customers and improve the quality of loans.