Inferences in Bayesian variable selection problems with large model spaces

摘要: An important aspect of Bayesian model selection is how to deal with huge spaces, since exhaustive enumeration all the models entertained unfeasible and inferences have be based on very small proportion visited. This case for variable problem, a moderate large number possible explanatory variables being considered in this paper. We review some strategies proposed literature argue that empirical frequencies via Markov Chain Monte Carlo sampling posterior distribution outperforms recently searching methods. give plausible yet simple explanation effect, showing estimators are unbiased. The results obtained two illustrative examples provide strong evidence favor our arguments.