E. Moreno Bas, F. A. Torres Ruiz, J. J. Serrano Pérez
A plethora of Bayes factors for variable selection, objective and subjective, can be found in the Bayesian literature of linear models. In this talk we deal with the question of what Bayes factor should be recommend to the users.
For moderate sample sizes the probabilistic properties of the Bayes factors as model selector are examined along with the probability that the objective and subjective Bayes factors select different models. The location of the sampling models for which there is discrepancy in the selection and the uncertainty of such a discrepancy are given.
For large sample sizes we present an analysis of the consistency of the Bayes factors when the dimension of the linear model p is either constant or grows at a rate p=O(n^{b}) for 0<b≤1.
Keywords: Key words: asymptotic, Bayes factors for nested models, complex linear models, g-priors, intrinsic priors, mixtures of g-priors, robust decisions.
Scheduled
GT21 Bayesian Inference II
June 7, 2022 3:30 PM
Audiovisual room
