Biometrika Advance Access originally published online on May 4, 2009
Biometrika 2009 96(3):497-512; doi:10.1093/biomet/asp017
Article |
Objective Bayesian model selection in Gaussian graphical models
Booth School of Business, University of Chicago, Chicago, Illinois 60637, U.S.A. carlos.carvalho{at}chicagobooth.edu
Department of Statistical Science, Duke University, Durham, North Carolina 27708, U.S.A. james{at}stat.duke.edu
Received for publication 1 September 2007. Revision received 1 November 2008.
This paper presents a default model-selection procedure for Gaussian graphical models that involves two new developments. First, we develop a default version of the hyper-inverse Wishart prior for restricted covariance matrices, called the hyper-inverse Wishart g-prior, and show how it corresponds to the implied fractional prior for selecting a graph using fractional Bayes factors. Second, we apply a class of priors that automatically handles the problem of multiple hypothesis testing. We demonstrate our methods on a variety of simulated examples, concluding with a real example analyzing covariation in mutual-fund returns. These studies reveal that the combined use of a multiplicity-correction prior on graphs and fractional Bayes factors for computing marginal likelihoods yields better performance than existing Bayesian methods.
Key Words: Bayesian model selection • Fractional Bayes factor • Gaussian graphical model • Hyper-inverse Wishart distribution • Multiple hypothesis testing
References
-
Atay-Kayis A., Massam H. A Monte Carlo method for computing the marginal likelihood in nondecomposable Gaussian graphical models. Biometrika (2005) 92:317–35.
Berger J. O., Pericchi L. The intrinsic Bayes factor for model selection and prediction. J. Am. Statist. Assoc. (1996) 91:109–22.[CrossRef][Web of Science]
Berger J. O., Pericchi L. Objective Bayesian methods for model selection: introduction and comparison. In: Model Selection (2001) Beachwood, OH: Institute of Mathematical Statistics. 135–207. Institute of Mathematical Statistics Lecture Notes – Monograph Series 38.
Carvalho C., Massam H., West M. Simulation of hyper-inverse Wishart distributions in graphical models. Biometrika (2007) 94:647–59.
Carvalho C., West M. Dynamic matrix-variate graphical models. Bayesian Anal. (2007) 2:69–96.[CrossRef]
Dawid A. P., Lauritzen S. L. Hyper-Markov laws in the statistical analysis of decomposable graphical models. Ann. Statist. (1993) 21:1272–317.[CrossRef]
Dobra A., Jones B., Hans C., Nevins J., West M. Sparse graphical models for exploring gene expression data. J. Mult. Anal. (2004) 90:196–212.[CrossRef]
Geisser S., Cornfield J. Posterior distributions for multivariate normal parameters. J. R. Statist. Soc. (1963) B25:368–76.
George E. I., Foster D. P. Calibration and empirical Bayes variable selection. Biometrika (2000) 87:731–47.
Giudici P. Learning in graphical Gaussian models. In: Bayesian Statistics 5—Berger J., Bernardo J., Dawid A., Smith A., eds. (1996) New York: Oxford University Press. 621–8.
Giudici P., Green P. J. Decomposable graphical Gaussian model determination. Biometrika (1999) 86:785–801.
Jefferys W., Berger J. Ockham's razor and Bayesian analysis. Am. Sci. (1992) 80:64–72.[Web of Science]
Jeffreys H. Theory of Probability (1961) 3rd ed. Oxford: Oxford University Press.
Jones B., Carvalho C., Dobra A., Hans C., Carter C., West M. Experiments in stochastic computation for high-dimensional graphical models. Statist. Sci. (2005) 20:388–400.[CrossRef]
Lauritzen S. L. Graphical Models (1996) Oxford: Clarendon Press.
Letac G., Massam H. Wishart distributions on decomposable graphs. Ann. Statist. (2007) 35:1278–1323.[CrossRef]
Liang F., Paulo R., Molina G., Clyde M., Berger J. Mixtures of g-priors for Bayesian variable selection. J. Am. Statist. Assoc. (2008) 103:410–23.[CrossRef][Web of Science]
Massam H., Neher E. Estimation and testing for lattice conditional independence models on euclidean jordan algebras. Ann. Statist. (1998) 26:1051–82.[CrossRef]
O'Hagan A. Fractional Bayes factors for model comparison. J. R. Statist. Soc. (1995) B 57:99–138.
Paulsen V., Power S., Smith R. Schur products and matrix completions. J. Funct. Anal. (1989) 85:151–78.[CrossRef]
Roverato A. Cholesky decomposition of a hyper-inverse Wishart matrix. Biometrika (2000) 87:99–112.
Scott J. G., Berger J. O. An exploration of aspects of Bayesian multiple testing. J. Statist. Plan. Infer. (2006) 136:2144–62.[CrossRef]
Scott J. G., Carvalho C. M. Feature-inclusion stochastic search for Gaussian graphical models. J. Comp. Graph. Statist. (2009) 17:790–808.[CrossRef]
Sun D., Berger J. O. Objective priors for the multivariate normal model. In: Bayesian Statistics 8—Bernardo J., Bayarri M., Berger J., Dawid A., Heckerman D., Smith A., West M., eds. (2009) Oxford: Oxford University Press.
Wermuth N. Linear recursive equations, covariance selection and path analysis. J. Am. Statist. Assoc. (1980) 75:963–72.[CrossRef][Web of Science]
Zellner A. On assessing prior distributions and Bayesian regression analysis with g-prior distributions. In: Bayesian Inference and Decision Techniques: Essays in Honor of Bruno de Finetti (1986) Amsterdam: North-Holland. 233–43.
Zellner A., Siow A. Posterior odds ratios for selected regression hypotheses. Bayesian Statistics: Proceedings of the First International Meeting held in Valencia (1980) Valencia: University Press.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||