Conditional kernel density estimation for some incomplete data models (Q1753141)

The authors propose a class of density estimators based on incomplete data. The method is to use a conditional kernel to construct the density estimator. Conditional kernel is defined as the expectation of a given kernel for the complete data conditioned on the observed data. The authors study asymptotic properties of such density estimators for several incomplete data models such as interval censoring model, convolution model, double censoring model and multiplicative censoring model. They observed that the asymptotic results of the proposed estimator do not depend on the choice of the kernel.