Censored data is a common problem in any kind of research. If censored data is ignored, the results could distort the facts. Therefore censored data should be considered. Quantile regression (QR) is a common way to investigate the possible relationsh...
Censored data is a common problem in any kind of research. If censored data is ignored, the results could distort the facts. Therefore censored data should be considered. Quantile regression (QR) is a common way to investigate the possible relationships between a covariate and a response variable . The quantile regression approach allows the analyst to estimate the functional dependence between variables for all portions of the conditional distribution of the response variable. In this article, we propose new nonparametric estimators of the quantile regression, which are the modified versions of double-kernel technique of Yu and Jones (1998) and the local logistic regression in Lee et al. (2006) under random censoring. We compare the new proposal with some existing methods. Those include the approach by Gannoun et al. (2007) and Ghouch and Van Keilegom (2009) based on the ‘check function’. The comparison is done by integrated squared error through a simulation study. We find that the modified version of the local logistic regression and the check function approach of Gannoun et al. (2007) performs better in most cases.