摘要
本文针对经典聚类分析和普通主成分聚类分析极端情形下的失效问题展开讨论,通过定义客观赋权的主成分距离为分类统计量,并以实证检验取得良好效果为依据,有效地解决了主成分聚类分析在极端情形下所不能揭示的问题。
A discussion about the invalidations of canonical cluster analysis and ordinary principal component cluster analysis under extreme situations is presented in this paper,by defining an objective weighted principal component distance as classification statistic,this paper solves the problems effectively which can＇t be revealed by principal component cluster analysis under extreme situations,and the empirical test proves its validation.
出处
《统计研究》
CSSCI
北大核心
2012年第11期84-87,共4页
Statistical Research
基金
国家社会科学基金项目"金融高频数据挖掘方法及应用研究"(11BTJ001)资助
关键词
分类
主成分聚类分析
极端情形
Classification
Principal component cluster analysis
Extreme situations
