摘要
基于核主成分分析方法,提出了复共线影响点的核主成分诊断以及剔除准则.数值试验表明,利用核主成分分析方法诊断剔除复共线影响点比用主成分方法更有效.
A new kernel principal component diagnosis method for muhicollinear influential points with the e-liminate criterion is put forward based on kernel principal component analysis. Numerical examples shows that eliminate muhicollinear influential points according to kernel principal component diagnosis method is more ef-fective then that of principal component diagnosis method.
出处
《鲁东大学学报(自然科学版)》
2012年第4期289-292,共4页
Journal of Ludong University:Natural Science Edition
关键词
复共线影响点
主成分分析
核主成分分析
:muhicollinear influential points
principal component analysis
kernel principal component analysis