Department of Statistics Unitmark
Dietrich College of Humanities and Social Sciences

Movies for the Visualization of MCMC Output

Publication Date

June, 2000

Publication Type

Tech Report

Author(s)

Nicole A. Lazar and Joseph B. Kadane

Abstract

Much work has been done on diagnostics for Markov Chain Monte Carlo analyses, although there are still many open questions. Reversible jump Markov Chain Monte Carlo, where the dimension of the model is also allowed to vary, poses special difficulties. We propose a movie as a dynamic way of assessing convergence and monitoring the behavior of the joint posterior distribution. This idea is demonstrated on a multiple changepoint problem, with data from a corbelled tomb (tholos) in Stylos, Crete.

(Revised 06/01, formerly Reversible Jump Markov Chain Monte Carlo: The Movie)