Influence in principal components analysis. (English) Zbl 0608.62068

Critchley, Frank

Biometrika 72, 627-636 (1985).

To detect outliers in data analyzed for principal components, three types of influence functions are derived based on perturbation parameters. Influence of these functions on the eigenvalues and eigenvectors of the covariance matrix is examined. The functions are compared among themselves and are contrasted with the regression case.

Reviewer: J.S.Murty

MSC:

62H25	Factor analysis and principal components; correspondence analysis
62F35	Robustness and adaptive procedures (parametric inference)

Keywords:

linear regression; outliers; principal components; influence functions; perturbation parameters; eigenvalues; eigenvectors; covariance matrix