Bayesian Missing Data Problems: EM, Data Augmentation and Non-iterative Computation.

摘要

This book is presented as a graduate text, as a reference for applied researchers, and even as a text for undergraduates and practitioners who seek to understand Monte Carlo simulation and Bayesian approaches to missing-data problems. Its key conceptual idea, motivated by fixed-point theorems in functional analysis, is to take the integral equation used by Tanner and Wong (1987) to motivate data augmentation and to make substitutions that give rise to what is termed the "inverse Bayes formula." The authors acknowledge that questions have been raised about the usefulness of such a framework but nonetheless press ahead. Specifically, using