Parameter Priors for Directed Acyclic Graphical Models and the Characterization of Several Probability Distributions

Abstract

We develop simple methods for constructing parameter priors for model choice among Directed Acyclic Graphical (DAG) models. In particular, we introduce several assumptions that permit the construction of parameter priors for a large number of DAG models from a small set of assessments. We then present a method for directly computing the marginal likelihood of every DAG model given a random sample with no missing observations. We apply this methodology to Gaussian DAG models which consist of a recursive set of linear regression models. We show that the only parameter prior for complete Gaussian DAG models that satisfies our assumptions is the normal-Wishart distribution. Our analysis is based on the following new characterization of the Wishart distribution: let $W$ be an $n\times n$, $n\ge 3$, positive-definite symmetric matrix of random variables and $f(W)$ be a pdf of $W$. Then, f$(W)$ is a Wishart distribution if and only if $W_{11}-W_{12}{W}_{22}^{-1}{W}_{12}^{\prime }$ is independent of $\{W_{12},W_{22}\}$ for every block partitioning $W_{11},W_{12},W_{12}^{\prime },W_{22}$ of $W$. Similar characterizations of the normal and normal-Wishart distributions are provided as well.

arXiv commentThis version has improved pointers to the literature. arXiv admin note: substantial text overlap with arXiv:1301.6697

References in corpus (2)

• A Transformational Characterization of Equivalent Bayesian Network Structures
• Learning Bayesian Networks: A Unification for Discrete and Gaussian Domains

Cited by in corpus (19)

• Learning Gaussian Graphical Models With Fractional Marginal Pseudo-likelihood
• Structure Learning of Probabilistic Graphical Models: A Comprehensive Survey
• Variational Causal Networks: Approximate Bayesian Inference over Causal Structures
• A review of Gaussian Markov models for conditional independence
• Extension of the Olkin and Rubin Characterization to the Wishart distribution on homogeneous cones
• Kummer and gamma laws through independences on trees - another parallel with the Matsumoto-Yor property
• Posterior Graph Selection and Estimation Consistency for High-dimensional Bayesian DAG Models
• Towards Scalable Bayesian Learning of Causal DAGs
• Complexity analysis of Bayesian learning of high-dimensional DAG models and their equivalence classes
• Structure learning in polynomial time: Greedy algorithms, Bregman information, and exponential families
• Bayesian Allocation Model: Inference by Sequential Monte Carlo for Nonnegative Tensor Factorizations and Topic Models using Polya Urns
• Retrospective-prospective symmetry in the likelihood and Bayesian analysis of case-control studies
• Structure learning of Bayesian networks involving cyclic structures
• Kshirsagar--Tan independence property of beta matrices and related characterizations
• A closed-form approach to Bayesian inference in tree-structured graphical models
• Estimating Bayesian networks for high-dimensional data with complex mean structure and random effects
• Learning Bayesian Networks from Ordinal Data
• Inference of Temporally Varying Bayesian Networks
• Objective Bayes Covariate-Adjusted Sparse Graphical Model Selection