ANALYSIS OF INFLUENTIAL OBSERVATIONS IN SEMIPARAMETRIC REGRESSION MODEL
In this article, we consider the semiparametric regression model, that consists parametric and nonparametric components simultaneouslyandexamineinfluentialobservationswhichhaveundue effects on estimators for this model. These observations are precisely detected by the major influence measures such as Cook’s distance, Hadi’s measure, etc. in linear regression model. The Cook’s distances for and in semiparametric regression model are expressed by Kim et al. [6] as functions of the residuals and leverages. Following the study of Kim et al. [6], we express a type of Hadi’s measure as functions of the residuals and leverages for the same model. We compare the performance of these measures as to detection of influential observations using real data and simulation. The results show that the performance of Hadi’s measure gives similar results with Cook’s distance for but better than Cook’s distances for and in the semiparametric regression model.
semiparametric regression, influential observations, Hadi’s measure, Cook’s distance.