Translator Disclaimer
June 2014 Local-Mass Preserving Prior Distributions for Nonparametric Bayesian Models
Juhee Lee, Steven N. MacEachern, Yiling Lu, Gordon B. Mills
Bayesian Anal. 9(2): 307-330 (June 2014). DOI: 10.1214/13-BA857

Abstract

We address the problem of prior specification for models involving the two-parameter Poisson-Dirichlet process. These models are sometimes partially subjectively specified and are always partially (or fully) specified by a rule. We develop prior distributions based on local mass preservation. The robustness of posterior inference to an arbitrary choice of overdispersion under the proposed and current priors is investigated. Two examples are provided to demonstrate the properties of the proposed priors. We focus on the three major types of inference: clustering of the parameters of interest, estimation and prediction. The new priors are found to provide more stable inference about clustering than traditional priors while showing few drawbacks. Furthermore, it is shown that more stable clustering results in more stable inference for estimation and prediction. We recommend the local-mass preserving priors as a replacement for the traditional priors.

Citation

Download Citation

Juhee Lee. Steven N. MacEachern. Yiling Lu. Gordon B. Mills. "Local-Mass Preserving Prior Distributions for Nonparametric Bayesian Models." Bayesian Anal. 9 (2) 307 - 330, June 2014. https://doi.org/10.1214/13-BA857

Information

Published: June 2014
First available in Project Euclid: 26 May 2014

zbMATH: 1327.62149
MathSciNet: MR3216998
Digital Object Identifier: 10.1214/13-BA857

Rights: Copyright © 2014 International Society for Bayesian Analysis

JOURNAL ARTICLE
24 PAGES


SHARE
Vol.9 • No. 2 • June 2014
Back to Top