Selection of GLM mixtures: a new criterion for clustering purpose

Abstract : Model-based clustering from finite mixtures of generalized linear models is a challenging issue which has undergone many recent developments. In practice, the model selection step is usually performed by using AIC or BIC penalized criteria. Though, simulations show that they tend to overestimate the actual dimension of the model. These evidence led us to consider a new criterion close to ICL, firstly introduced in Baudry (2009). Its definition requires to introduce a contrast embedding an entropic term: using concentration inequalities, we derive key properties about the convergence of the associated M-estimator. The consistency of the corresponding classification criterion then follows depending on some classical requirements on the penalty term. Finally a simulation study enables to corroborate our theoretical results, and shows the effectiveness of the method in a clustering perspective.
Type de document :
Pré-publication, Document de travail
Liste complète des métadonnées

Littérature citée [35 références]  Voir  Masquer  Télécharger
Contributeur : Olivier Lopez <>
Soumis le : mardi 11 mars 2014 - 13:53:11
Dernière modification le : mardi 30 mai 2017 - 01:17:30
Document(s) archivé(s) le : mercredi 11 juin 2014 - 11:36:23


Fichiers produits par l'(les) auteur(s)


  • HAL Id : hal-00957880, version 1



Olivier Lopez, Milhaud Xavier. Selection of GLM mixtures: a new criterion for clustering purpose. 2014. 〈hal-00957880〉



Consultations de la notice


Téléchargements de fichiers