摘要
P2P是一种新型的互联网借贷模式,以其高效、便利性迅速扩张。信息的不对称,使得P2P网络借贷具有较高的风险。准确度量借款人的信用风险是P2P网贷平台维护正常经营、降低信用风险以及保护投资者利益的重要对策。本文引入Lasso算法并与Logistic模型相结合,根据"人人贷"平台上的借贷数据,建立信用风险度量模型。通过实证发现:信用等级、年龄、受教育程度、逾期次数以及成功借款笔数对信用风险有显著的影响,且Lasso-Logistic模型有较高的预测准确率。
P2P is a new type of Internet borrowing and lending model,which expands rapidly with its high efficiency and convenience.Because of the asymmetric information,P2P Internet loan has a high credit risk.Accurate prediction of the borrowers'credit risk is an important way for P2P platform to maintain normal operation,reduce credit risk and protect investors'interest.In this paper,the combination of Lasso algorithm and Logistic model is introduced to establish a credit risk prediction model based on the borrowing and lending data from“Renrendai”platform.The empirical result shows that the borrower’s credit level,age,education degree,times of overdue repayment and times of successful borrowing have a significant impact on the credit risk,and Lasso-Logistic model has a higher prediction accuracy.
作者
邹明芮
ZOU Mingrui(School of Finance,Guangdong University of Foreign Studies,Guangzhou 510000,China)
出处
《长春大学学报》
2018年第3期22-26,共5页
Journal of Changchun University
基金
广东省教育厅"十三五"规划项目(粤教师函[2017]145号)
作者简介
邹明芮(1991-),女,山东淄博人,讲师,硕士,主要从事互联网金融研究。
