zbMATH — the first resource for mathematics

Objective Bayes factors for Gaussian directed acyclic graphical models. (English) Zbl 1253.62015
Summary: We propose an objective Bayesian method for the comparison of all Gaussian directed acyclic graphical models defined on a given set of variables. The method, which is based on the notion of fractional Bayes factors (BFs), requires a single default (typically improper) prior on the space of unconstrained covariance matrices, together with a prior sample size hyper-parameter, which can be set to its minimal value. We show that our approach produces genuine BFs. The implied prior on the concentration matrix of any complete graph is a data-dependent Wishart distribution, and this in turn guarantees that Markov equivalent graphs are scored with the same marginal likelihood. We specialize our results to the smaller class of Gaussian decomposable undirected graphical models and show that in this case they coincide with those recently obtained using limiting versions of hyper-inverse Wishart distributions as priors on the graph-constrained covariance matrices.

62F15 Bayesian inference
05C20 Directed graphs (digraphs), tournaments
62H10 Multivariate distribution of statistics
05C90 Applications of graph theory
Full Text: DOI
[1] Andersson, On the Markov equivalence of chain graphs, undirected graphs, and acyclic digraphs, Scand. J. Statist. 24 pp 81– (1997) · Zbl 0918.60050
[2] Berger, The intrinsic Bayes factor for model selection and prediction, J. Amer. Statist. Assoc. 91 pp 109– (1996) · Zbl 0870.62021
[3] Berger, Model selection pp 135– (2001)
[4] Carvalho, Objective Bayesian model selection in Gaussian graphical models, Biometrika 96 pp 497– (2009) · Zbl 1170.62020
[5] Consonni, Bayesian statistics 9-Proceedings of the ninth Valencia international meeting pp 119– (2011)
[6] Consonni, Conjugate priors for exponential families having quadratic variance functions, J. Amer. Statist. Assoc. 87 pp 1123– (1992) · Zbl 0764.62027
[7] Cowell, Probabilistic networks and expert systems (1999) · Zbl 0937.68121
[8] Dawid, Highly structured stochastic systems pp 45– (2003)
[9] Dawid, Hyper Markov laws in the statistical analysis of decomposable graphical models, Ann. Statist. 21 pp 1272– (1993) · Zbl 0815.62038
[10] DeGroot, Optimal statistical decisions (1970)
[11] Diaconis, Conjugate priors for exponential families, Ann. Statist. 7 pp 269– (1979) · Zbl 0405.62011
[12] Friedman, Being Bayesian about network structure-A Bayesian approach to structure discovery in Bayesian networks, Machine Learning 50 pp 95– (2003) · Zbl 1033.68104
[13] Geiger, Parameter priors for directed acyclic graphical models and the characterization of several probability distributions, Ann. Statist. 30 pp 1412– (2002) · Zbl 1016.62064
[14] Gutiérrez-Peña, Conjugate parameterizations for natural exponential families, J. Amer. Statist. Assoc 90 pp 1347– (1995) · Zbl 0868.62029
[15] Johnson, On the use of non-local prior densities in Bayesian hypothesis tests., J. Roy. Statist. Soc. Ser. B Statist. Methodol. 72 pp 143– (2010)
[16] Lauritzen, Graphical models (1996)
[17] Lauritzen, Complex stochastic systems pp 63– (2001)
[18] Letac, Wishart distributions for decomposable graphs, Ann. Statist. 35 pp 1278– (2007) · Zbl 1194.62078
[19] Moreno, L1-statistical procedures and related topics pp 257– (1997)
[20] O’Hagan, Fractional Bayes factors for model comparison, J. Roy. Statist. Soc. Ser. B Statist. Methodol. 57 pp 99– (1995)
[21] Perez, Expected-posterior prior distributions for model selection, Biometrika 89 pp 491– (2002) · Zbl 1036.62026
[22] Pericchi, Bayesian thinking: modeling and computation pp 115– (2005)
[23] Press, Applied multivariate analysis: using Bayesian and frequentist methods of inference (1982)
[24] Scott, Feature-inclusion stochastic search for Gaussian graphical models, J. Comput. Graph. Statist. 17 pp 790– (2009)
[25] Whittaker, Graphical models in applied multivariate statistics (1990) · Zbl 0732.62056
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.