Observed Range Maximum Likelihood Estimation

TL;DR

This paper extends maximum likelihood estimation techniques to multivariate censored data, proposing kernel-based methods and applying them to multinomial and contingency table analysis with censoring.

Contribution

It introduces a generalized observed range MLE, a kernel density estimator, and methods for censored multinomial and contingency table data analysis.

Findings

Extended MLE to multivariate censored data

Developed a kernel density estimator for censored data

Applied methods to multinomial and contingency tables

Abstract

The idea of maximizing the likelihood of the observed range for a set of jointly realized counts has been employed in a variety of contexts. The applicability of the MLE introduced in [1] has been extended to the general case of a multivariate sample containing interval censored outcomes. In addition, a kernel density estimator and a related score function have been proposed leading to the construction of a modified Nadaraya-Watson regression estimator. Finally, the author has treated the problems of estimating the parameters of a mutinomial distribution and the analysis of contingency tables in the presence of censoring.