ANALISIS REGRESI SEMIPARAMETRIK PADA KASUS HILANGNYA RESPON

Abstract

In the specific cases of experiment, not all data (response) may be available, which is called missing response cases. It’s appear for various reasons.  For the existing problem, inference statistics cannot be applied directly.  The aim of this research is to consider about certain method to impute the missing response which is related to semiparametric regression, as a goodness of fit measurement of the used method, suppose an estimator  which is compared to the mean of complete response, then consider asymptotic distribution, consistency and efficiency of parametrics component estimator.

By using Kernel approximation, the resulted of nonparametrics estimator and by least square method, the resulted parametric component .The application to minimum temperature’s data in 56 cities at USA, estimator value of  for several confidence interval tend to be similar to the mean value of complete response.