Thumbnail - No thumbnail

Jeffreys priors for mixture estimation: properties and alternatives

Grazian, Clara; Robert, Christian P. (2018), Jeffreys priors for mixture estimation: properties and alternatives, Computational Statistics & Data Analysis, 121, p. 149-163. 10.1016/j.csda.2017.12.005

Type
Article accepté pour publication ou publié
Lien vers un document non conservé dans cette base
https://arxiv.org/abs/1511.03145
Date
2018
Nom de la revue
Computational Statistics & Data Analysis
Volume
121
Pages
149-163
Identifiant publication
Métadonnées
Afficher la notice complète
Auteur(s)
Grazian, Clara

Robert, Christian P.
Résumé (EN)
While Jeffreys priors usually are well-defined for the parameters of mixtures of distributions, they are not available in closed form. Furthermore, they often are improper priors. Hence, they have never been used to draw inference on the mixture parameters. We study in this paper the implementation and the properties of Jeffreys priors in several mixture settings, show that the associated posterior distributions most often are improper, and then propose a non-informative alternative for the analysis of mixtures.
Mots-clés
Noninformative prior; mixture of distributions; Bayesian analysis; Dirichlet prior; improper prior; improper posterior; labelswitching