## Bayesian model choice based on Monte Carlo estimates of posterior model probabilities.(English)Zbl 1330.62115

Summary: A range of approximate methods have been proposed for model choice based on Bayesian principles, given the problems involved in multiple integration in multi-parameter problems. Formal Bayesian model assessment is based on prior model probabilities $$P(M=j)$$ and posterior model probabilities $$P(M=j|Y)$$ after observing the data. An approach is outlined here that produces posterior model probabilities and hence Bayes factor estimates but not marginal likelihoods. It uses a Monte Carlo approximation based on independent MCMC sampling of two or more different models. While parallel sampling of the models is not necessary, such a form of sampling facilitates model averaging and assessing the impact of individual observations on the overall estimated Bayes factor. Three worked examples used before in model choice studies illustrate application of the method.

### MSC:

 62F15 Bayesian inference 62-07 Data analysis (statistics) (MSC2010) 65C40 Numerical analysis or methods applied to Markov chains 62P20 Applications of statistics to economics

BayesDA
Full Text:

### References:

 [1] Carlin, B.; Chib, S., Bayesian model choice via Markov chain Monte Carlo methods, J. roy. statist. soc. ser. B, 57, 473-484, (1995) · Zbl 0827.62027 [2] Chen, M.; Shao, Q.; Ibrahim, J., Monte Carlo methods in Bayesian computation, (2000), Springer New York · Zbl 0949.65005 [3] Chib, S., Marginal likelihood from the Gibbs output, J. amer. statist. assoc., 90, 1313-1321, (1995) · Zbl 0868.62027 [4] Congdon, P., 2004. Bayesian predictive model comparison via parallel sampling. Computational Statistics and Data Analysis, in press. · Zbl 1429.62093 [5] Dellaportas, P.; Forster, J.; Ntzoufras, I., On Bayesian model and variable selection using MCMC, Statist. comput., 12, 27-36, (2002) · Zbl 1247.62086 [6] Gelfand, A.; Dey, D., Bayesian model choiceasymptotics and exact calculations, J. roy. statist. soc. ser. B, 56, 501-514, (1994) · Zbl 0800.62170 [7] Gelman, A.; Meng, X., Simulating normalizing constantsfrom importance sampling to bridge sampling to path sampling, Statist. sci., 13, 163-185, (1998) · Zbl 0966.65004 [8] Gelman, A.; Carlin, J.; Stern, H.; Rubin, D., Bayesian data analysis, (1995), Chapman & Hall London [9] Godsill, S., On the relationship between Markov chain Monte Carlo methods for model uncertainty, J. comput. graphical statist., 10, 230-248, (2001) [10] Green, P., Reversible jump Markov chain Monte Carlo computation and Bayesian model determination, Biometrika, 82, 711-732, (1995) · Zbl 0861.62023 [11] Green, P., Trans-dimensional Markov chain Monte Carlo, (), 179-198 [12] Green, P., O’Hagan, A., 2000. Model choice with MCMC on product spaces without using pseudo priors. Technical Report, Department of Statistics, University of Nottingham. [13] Kadane, J.; Lazar, N., Methods and criteria for model selection, J. amer. statist. assoc., 99, 279-290, (2004) · Zbl 1089.62501 [14] Moreno, E.; Liseo, B., A default Bayesian test for the number of components in a mixture, J. statist. plann. inference, 111, 129-142, (2003) · Zbl 1033.62025 [15] Newton, M.; Raftery, A., Approximate Bayesian inference with the weighted likelihood bootstrap, J. roy. statist. soc. ser. B, 56, 3-48, (1994) · Zbl 0788.62026 [16] Pettit, L.; Young, K., Measuring the effect of observations on Bayes factors, Biometrika, 77, 455-466, (1990) [17] Raftery, A., Approximate Bayes factors and accounting for model uncertainty in generalised linear models, Biometrika, 83, 251-266, (1996) · Zbl 0864.62049 [18] Scott, S., Bayesian methods for hidden Markov modelsrecursive computing in the 21st century, J. amer. statist. assoc., 97, 337-351, (2002) · Zbl 1073.65503 [19] Song, X.; Lee, S., A Bayesian model selection method with applications, Computational statistics and data analysis, 40, 539-557, (2002) · Zbl 0993.62026 [20] Wasserman, L., Bayesian model selection and model averaging, J. math. psychol., 44, 92-107, (2000) · Zbl 0946.62032
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.