信用评级的新方法——多元自适应回归样条在民营企业信用评级中的应用  

New Method of Credit Rating——The Application of Private Enterprises Credit Rating Based on Multivariate Adaptive Regression Splines

作　　者：孟庆福[1] 杜元园[1] 曲华锋[1] 

机构地区：[1]吉林大学商学院,吉林长春130012

出　　处：《广东金融学院学报》2011年第5期65-76,共12页Journal of Guangdong University of Finance

基　　金：国家社会科学基金项目(05BJL022);教育部人文社会科学重点研究基地2006年度重大研究项目(06JJD790012);吉林大学"985工程"项目

摘　　要：通过建立多元自适应回归样条模型,对民营企业是否违约进行两等级划分,结果表明总体正确率达95.9%,对第二类错误的判别正确率达80%,均高于多元判别分析模型和Logistic回归模型。因此可以证明,多元自适应回归样条模型应用于判别民营企业的信用等级具有较好的实践价值。By setting up the multivariate adaptive regression splines model,we can discriminate the private enterprises into two grades.The result showed that the integer accuracy was 95.9%,the accuracy of the second type error was 80%.The accuracy was better than using multivariate discriminate model and Logistic model.This proved that it was valuable for practice to use multivariate adaptive regression splines model to study the credit rating of private enterprises.

关 键 词：民营企业 多元自适应回归样条 信用评级 

分 类 号：F830.56[经济管理—金融学]