Local influence in principal component analysis: relationship between the local influence and influence function approaches revisited

摘要: Abstract Influence analysis is developed on the basis of Cook's local influence in its original form, i.e., using likelihood displacement as criterion function for studying jointly well singly influential observations principal component (PCA). It found that derived directions are equivalent to standardized vectors scores obtained by PCA with metric V− functions parameters θ , where V a consistent estimate asymptotic covariance matrix and superscript (−) indicates g-inverse. special case equivalence discussed Tanaka Zhang (Comp. Statist. Data Anal. 32 (1999) 197) general statistical modeling. This broadens range theory holds. A numerical example given illustrate performance proposed compare results those previous studies.