基于变量双重检验的Fisher信用风险度量模型  

Fisher Discriminant Model of Credit Risk Based on Double Test to the Independent Variables

作　　者：何树红[1] 王善民[1] 

机构地区：[1]云南大学数学系,云南昆明650091

出　　处：《系统工程》2007年第8期15-20,共6页Systems Engineering

基　　金：国家自然科学基金资助项目(10561009);云南大学"中青年骨干教师培养计划"专项经费资助项目

摘　　要：为提高模型入选样本变量的准确性与稳定性,本文采用附加信息检验和多重共线性检验选择自变量,并应用Fisher判别原理建立信用风险度量函数。实证结果显示,所建立的模型对全部246个样本的误判率仅为6.91%,准确率为93.09%;而对风险样本的误判率为4.55%,正确辨别精度高达95.45%。Applying additional test and multilinear test, independent variables are selected, and then Fisher diseriminant function and credit-risk model are established. The result shows that, to all samples, the error rate of the model discriminent is 6.91 % ,while the accuracy rate is 93.09%. As to venture sample,the error rate is merely 4.55% ,and the accuracy is up to 95.45%.

关 键 词：信用风险 附加信息检验 多重共线性检验 FISHER判别 

分 类 号：F830[经济管理—金融学]