ABSTRACT
The chapter explores a number of examples with missing, truncated and censored data when the problem is to estimate an underlying probability density. Nonparametric curve estimation allows one to analyze data without assuming t&he shape of an estimated curve. Methods of nonparametric estimation are well developed for the case of directly observed data, much less is known for the case of missing and modified data. The chapter explains the problem of nonparametric regression with missing data via real and simulated examples. Histogram is a nonparametric density estimator because no assumption about shape of an underlying density is made. The chapter argues that different estimation procedures are needed for cases where either responses or predictors are missed. It highlights the problem of estimation of nuisance functions and offers exercises that allow the reader to review basics of probability and statistics. The chapter also presents an overview of the key concepts discussed in this book.
