一种判别模型的择优方法  

A selection method of discriminant model

作　　者：黄利文[1] 

机构地区：[1]泉州师范学院数学与计算机学院,福建泉州362000

出　　处：《江西理工大学学报》2013年第1期96-99,共4页Journal of Jiangxi University of Science and Technology

基　　金：泉州师范学院科技项目(2012KJ016)

摘　　要：文中对判别模型中的回代正确率评价准则进行了推广.首先对判别模型的预判能力和误判程度进行度量,然后根据层次分析法的思路以模型的预判能力、稳健性及误判程度为准则层提出了判别模型的择优方法,并以该方法选择判别模型.实例表明,该方法考虑了模型的预判能力、稳健性和误判程度,解决了回代正确率的评价准则在选择判别模型上的单一性,即仅考虑模型的预判能力.同时,多个模型的回代正确率相同时,不易选择较优的模型,文中的方法为选择更好的判别模型提供了依据.This paper extends the criterion of the right ratio of discriminant model. Firstly, this paper measures the anticipation ability and misclassification degree; then a selection method of discriminant model is presented by the rule of anticipation ability, robustness and misclassification according to the method of Analytic Hierarchy Process （AHP）, by which the discriminant model is selected. The example shows that this method considers the anticipation ability, robustness and misclassification ＂degree of discriminant model and solves the problems of the right ratio of discriminant model which only considers the anticipation ability of discriminant model. Besides, when the right ratios of several models are equal and it is not easy to get the better discriminant model, this method provides the basis to select a better discriminant model.

关 键 词：判别模型 误判 择优方法 层次分析法 

分 类 号：O212[理学—概率论与数理统计]