Nonparametric Regression Estimation with Missing Censoring Indicators

Nonparametric regression estimation has been studied intensively for the censored data. However, in some practical applications, some censoring indicators may be missing because of various reasons. In this paper, we propose two kernel estimators for the regression function when the censoring indicator is missing at random. The strong uniform convergence rates and the asymptotic normality of the estimators are established. Some simulations are carried out to assess the finite sample performances of the proposed methods.