关键词: 分类, 主成分聚类分析, 极端情形

Abstract: 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.

Key words: Classification, Principal Component Cluster Analysis, Extreme Situations