摘要
针对传统主成分在处理非线性问题上的不足,阐述了传统方法在数据无量纲化中"中心标准化"的缺点和处理"线性"数据时的缺陷,给出了数据无量纲化和处理"非线性"数据时的改进方法,并建立了一种基于"对数中心化"的非线性主成分分析和聚类分析的新的综合评价方法。实验表明,该方法能有效地处理非线性数据。
Against the weaknesses of traditional principal component analysis when it is used to solve nonlinear problems, this paper points out its shortage of “centralized criterion in data un - dimensionalization” and its drawbacks on processing “non - linear” data firstly, puts forward improvement secondly, and then builds up a new synthetic evaluation based on non - linear principal component analysis according to “centralized logarithm” and cluster analysis in the end. The research result shows that the new method can effectually deal with non- linear data.
出处
《统计与信息论坛》
CSSCI
2008年第2期37-41,46,共6页
Journal of Statistics and Information
关键词
主成分分析
中心标准化
均值化
对数中心化
聚类分析
principal component analysis
centralized criterion
equalization
centralized logarithm
cluster analysis
