摘要
This paper proposes a confidence interval for the number of important principal components in principal component analysis.An important component is defined as a principal component whose value is close to the value of the largest component.More specifically,a component λ i is called important if λ i> λ 1 is sufficiently close to 1.In the situation when several components are close to the top component and their variances are almost inseparable,our procedure will include all those “important” components while the usual criterion might miss some of them.
This paper proposes a confidence interval for the number of important principal components in principal component analysis.An important component is defined as a principal component whose value is close to the value of the largest component.More specifically,a component λ i is called important if λ i> λ 1 is sufficiently close to 1.In the situation when several components are close to the top component and their variances are almost inseparable,our procedure will include all those “important” components while the usual criterion might miss some of them.
出处
《统计研究》
CSSCI
北大核心
2005年第6期53-58,共6页
Statistical Research
