BayesVarSel: Bayesian Variable selection in Linear Models

Within the context of the lineal model, this package provides tools for the analysis of the variable selection problem from a Bayesian perspective. Special emphasis is placed on providing simple and intuitive methods to explore and synthesize the results. BayesVarSel allows the calculations to be performed either exactly (sequential or parallel computation) or heuristically, using a Gibbs sampling algorithm studied in Garcia-Donato and Martinez-Beneito (2012). The default implementation takes advantage of a closed-form expression for the posterior probabilities that the prior proposed in Bayarri, Berger, Forte and Garcia-Donato (2012) produces. Nevertheless, other alternative priors, like Zellner (1986) g-prior, Zellner-Siow (1980,1984) or Liang, Paulo, Molina, Clyde and Berger (2008) can be used.

Downloads: