| 多类母体的广义距离判别及其应用 |
| |
| 引用本文： | 王学仁,张进.多类母体的广义距离判别及其应用[J].云南大学学报(自然科学版),1993,15(3):191-199. |
| |
| 作者姓名： | 王学仁 张进 |
| |
| 作者单位： | 云南省应用数学研究所 云南大学统计系
(王学仁)，云南省应用数学研究所 云南大学统计系(张进) |
| |
| 摘 要： | 关于多类母体的判别问题,当母体间的差异主要不是表现在均值上而是“方差”上时,常用的判别方法都会失效。为此,作者引入了广义的距离判别法。此法不仅实用范围广,而且显著地提高了判别效率。此外,在正态假设下,它等价子Bayes的二次判别,因而是最优判别。
|
| 关 键 词： | 判别分析 距离判别法 多元统计分析 |
The Discrimination Based on Generalized Mahalanobis/ Dis- tance and Its Applications |
| |
| Wang Xueren Zhang Jin.The Discrimination Based on Generalized Mahalanobis/ Dis- tance and Its Applications[J].Journal of Yunnan University(Natural Sciences),1993,15(3):191-199. |
| |
| Authors: | Wang Xueren Zhang Jin |
| |
| Abstract: | In discriminant analysis, triditional methods of discrimination are invalid when the differences among populations lie mainly in their deviations instead of their meams. So, we introduce the discrimination based on generalized Mahalanobis' distance, which is effective. Under conditions of normal populations, this metihod is optimal because it is equivalant to Bayes' gudratic discrimination. |
| |
| Keywords: | Discriminant analysis Mahalanobis' distance Probability of misclassification |
| 本文献已被 CNKI 维普 等数据库收录！ |
|
