Ano: 2014
Código: WPE – 341
Autores/Pesquisadores:
Abstract:
In this review paper we revisit several of the existing schemes that approximate predictive densities and, consequently, Bayes factors. We also present the reversible jump MCMC scheme, which can be thought of as an MCMC scheme over the space of models. These approaches are applied to select the number of common factors in the basic normal linear factor model, which is a high pro le example within the psychometrics community.