Semıparametrıc Regressıon Estımates Based on Some Transformatıon Technıques for Rıght-Censored Data

Abstract

In this paper, we introduce three different data transformation approaches such as synthetic data transformation ([1]; [2]; [3]), Kaplan-Meier weights ([4]; [5]; [6]) and k-nearest neighbor (kNN) imputation method ([7]) which are commonly used in censored data applications. The aforementioned approaches are particularly useful when one deals with censored data. The key idea expressed here is to find the smoothing spline estimates for the parametric and nonparametric components of a semiparametric regression model with right censored data. The estimation is then carried out based on the modified (or transformed) data set obtained via these transformation techniques. In order to compare the outcomes of three approaches in semi-parametric regression setting, we carried out a simulation study. According to the results of the simulation, it can be said that the Kaplan-Meier weights has been very successful in dealing with censored observations.