Abstract:
|
Bayesian hierarchical models are increasingly popular tools for analyzing complex data sets. Unfortunately, traditional criteria for assessing adequacy of a single model and comparing alternative models, such as cross-validation sums of squares, are inappropriate for non-standard data structures. More flexible cross-validation criteria, such as predictive densities, facilitate effective evaluations, but do so at the expense of introducing computational difficulties. This paper considers Markov Chain Monte Carlo calculations of Bayesian predictive densities for vector measurements subject to differential component-wise censoring. It discusses computational obstacles resulting from both the multivariate and incomplete nature of the data, and suggests approaches for reducing Monte Carlo variability and overall computational burden. It demonstrates the value of the proposed methods in the context of comparing alternative models for joint distributions of contaminant concentration measurements.
|