Estimation and selection for the latent block model on categorical data

摘要: This paper deals with estimation and model selection in the Latent Block Model (LBM) for categorical data. First, after providing sufficient conditions ensuring identifiability of this model, we generalise procedures criteria derived binary Secondly, develop Bayesian inference through Gibbs sampling a well calibrated non informative prior distribution, order to get MAP estimator: is proved avoid traps encountered by LBM maximum likelihood methodology. Then are presented. In particular an exact expression integrated completed criterion requiring no asymptotic approximation derived. Finally numerical experiments on both simulated real data sets highlight appeal proposed procedures.