On choosing a mixture model for clustering
Résumé
Two methods for both clustering data and choosing a mixture model are proposed. First, the unknown clusters are assessed. Then, the likelihood conditional to these clusters is written as the product of likelihoods from each cluster. AIC and BIC type-approximations are then applied, and the resulting criteria turn out to be the sum of the AIC or BIC relative to each cluster. The performances of our methods are evaluated on real data examples and numerical simulations.
Origine : Fichiers produits par l'(les) auteur(s)