A conjugate prior for discrete hierarchical log-linear models

DOI10.1214/08-AOS669zbMATH Open1369.62048arXiv0711.1609OpenAlexW3102652972MaRDI QIDQ1043747FDOQ1043747

Authors: Hélène Massam, Jinnan Liu, Adrian Dobra 

Publication date: 9 December 2009

Published in: The Annals of Statistics (Search for Journal in Brave)

Abstract: In Bayesian analysis of multi-way contingency tables, the selection of a prior distribution for either the log-linear parameters or the cell probabilities parameters is a major challenge. In this paper, we define a flexible family of conjugate priors for the wide class of discrete hierarchical log-linear models, which includes the class of graphical models. These priors are defined as the Diaconis--Ylvisaker conjugate priors on the log-linear parameters subject to "baseline constraints" under multinomial sampling. We also derive the induced prior on the cell probabilities and show that the induced prior is a generalization of the hyper Dirichlet prior. We show that this prior has several desirable properties and illustrate its usefulness by identifying the most probable decomposable, graphical and hierarchical log-linear models for a six-way contingency table.

Full work available at URL: https://arxiv.org/abs/0711.1609

Recommendations

• Prior induction in log-linear models for general contingency table analysis.
• Bayesian selection of log‐linear models
• scientific article; zbMATH DE number 218412
• Empirical Bayes analysis of log-linear models for contingency tables
• Bayesian inference for Poisson and multinomial log-linear models

zbMATH Keywords

model selectionconjugate priorcontingency tableshierarchical log-linear modelshyper Dirichlethyper Markov property

Mathematics Subject Classification ID

Exact distribution theory in statistics (62E15) Bayesian inference (62F15) Contingency tables (62H17)

Cites Work

• Conjugate priors for exponential families
• Hyper Markov laws in the statistical analysis of decomposable graphical models
• Model uncertainty
• Title not available (Why is that?)
• MCMC model determination for discrete graphical models
• A fast procedure for model search in multidimensional contingency tables
• Accurate Approximations for Posterior Moments and Marginal Densities
• Title not available (Why is that?)
• Title not available (Why is that?)
• Model Selection and Accounting for Model Uncertainty in Graphical Models Using Occam's Window
• Bayesian methods for estimation of the size of a closed population
• Title not available (Why is that?)
• Bayesian Graphical Models for Discrete Data
• Markov chain Monte Carlo model determination for hierarchical and graphical log-linear models
• The mode oriented stochastic search (MOSS) algorithm for log-linear models with conjugate priors
• Markov fields and log-linear interaction models for contingency tables
• Additive and multiplicative models and interactions
• Prior induction in log-linear models for general contingency table analysis.
• Conjugate Parameterizations for Natural Exponential Families
• Incorporating Prior Information into the Analysis of Contingency Tables
• Bounds for cell entries in contingency tables given marginal totals and decomposable graphs
• Conjugate Priors for Exponential Families Having Quadratic Variance Functions
• Reference priors for discrete graphical models
• Title not available (Why is that?)

Cited In (19)

• Simplex factor models for multivariate unordered categorical data
• Specification of hyper-parameters for normal prior distributions induced on log-linear parame\-ters
• Optimal Gaussian approximations to the posterior for log-linear models with Diaconis-Ylvisaker priors
• A new prior for discrete DAG models with a restricted set of directions
• Parametrizations and reference priors for multinomial decomposable graphical models
• Modeling association in microbial communities with clique loglinear models
• Book Reviews
• Bayesian inference for Poisson and multinomial log-linear models
• Loglinear model selection and human mobility
• A Bayesian graphical model for genome-wide association studies (GWAS)
• Putting background information about relative risks into conjugate prior distributions
• Bayes factors and the geometry of discrete hierarchical loglinear models
• Maximum likelihood estimation in log-linear models
• Conjugate and conditional conjugate Bayesian analysis of discrete graphical models of marginal independence
• The mode oriented stochastic search (MOSS) algorithm for log-linear models with conjugate priors
• Composite mixture of log-linear models with application to psychiatric studies
• Bayesian learning of weakly structural Markov graph laws using sequential Monte Carlo methods
• Fully Bayesian binary Markov random field models: prior specification and posterior simulation
• Prior induction in log-linear models for general contingency table analysis.

This page was built for publication: A conjugate prior for discrete hierarchical log-linear models

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1043747)