Ioannis Ntzoufras

Publications Page

Computation for Intrinsic Variable Selection

in Normal Regression Models via Expected-Posterior Prior

Fouskakis, D. and Ntzoufras, I. (2013)

Statistics and Computing, 23, 491–499.

SYNOPSIS

In this paper we focus on the variable selection problem in normal regression models, using the expected-posterior prior methodology. We provide a straightforward MCMC scheme for the derivation of the posterior distribution, as well as Monte Carlo estimates for the computation of the marginal likelihood and posterior model probabilities. Additionally, for large model spaces, a model search algorithm based on MC³ is constructed. The proposed methodology is implemented in two real life examples, already used in the relevant literature of objective variable selection. In both illustrated examples, uncertainty over different training samples is also considered

 

Keywords: Bayesian variable selection; Expected posterior priors; Imaginary data; Intrinsic priors; Jeffreys prior; Objective model selection methods; Normal regression models.

Download:

• First version: 15/7/2011available here.

• Finally published version: [16/1/2012] available here: Fouskakis, D. and Ntzoufras, I. (2013). Computation for Intrinsic Variable Selection in Normal Regression Models via Expected-Posterior Prior Statistics and Computing, 23, 491–499.

---

All Contents Copyright.
Last revised: 13/12/2013