Skip to main content

Research Repository

Advanced Search

Variations of power-expected-posterior priors in normal regression models

Fouskakis, Dimitris; Ntzoufras, Ioannis; Perrakis, Konstantinos

Variations of power-expected-posterior priors in normal regression models Thumbnail


Authors

Dimitris Fouskakis

Ioannis Ntzoufras



Abstract

The power-expected-posterior (PEP) prior is an objective prior for Gaussian linear models, which leads to consistent model selection inference, under the M-closed scenario, and tends to favour parsimonious models. Recently, two new forms of the PEP prior were proposed which generalize its applicability to a wider range of models. The properties of these two PEP variants within the context of the normal linear model are examined thoroughly, focusing on the prior dispersion and on the consistency of the induced model selection procedure. Results show that both PEP variants have larger variances than the unit-information -prior and that they are M-closed consistent as the limiting behaviour of the corresponding marginal likelihoods matches that of the BIC. The consistency under the M-open case, using three different model misspecification scenarios is further investigated.

Citation

Fouskakis, D., Ntzoufras, I., & Perrakis, K. (2020). Variations of power-expected-posterior priors in normal regression models. Computational Statistics & Data Analysis, 143, Article 106836. https://doi.org/10.1016/j.csda.2019.106836

Journal Article Type Article
Acceptance Date Sep 4, 2019
Online Publication Date Sep 12, 2019
Publication Date Mar 1, 2020
Deposit Date Sep 26, 2019
Publicly Available Date Sep 12, 2020
Journal Computational Statistics & Data Analysis
Print ISSN 0167-9473
Publisher Elsevier
Peer Reviewed Peer Reviewed
Volume 143
Article Number 106836
DOI https://doi.org/10.1016/j.csda.2019.106836
Related Public URLs https://arxiv.org/pdf/1609.06926.pdf

Files





You might also like



Downloadable Citations