Assessment of anomalous observations in liu estimator

Abstract

Liu linear regression model building analysis is frequently applied for fitting model to near mullicollinearity data sets. When atypical observations exist in a data set, they may exert undue influence on the result of the analysis. This paper studies an assessment of the minor perturbation of the Liu estimator in the ridge type linear regression model using Cook's (1986) method to detect anomalous observations in the data set when mean squared error is known or unknown, and perturbation of individual explanatory variable. Example based on Longley data are used for illustration.