Modelo Weibull multivariado para dados dependentes

Abstract

In this work a multivariate model was proposed for the data that exhibit dependence. The model was obtained by marginalizing of Weibull G-exponentialised distribution conditioned to a latent variable and/or random field formed by a mixture of am-stable distributions. The latent variables were incorporated into the model to capture the dependency of interest. Some properties of this new model are presented and procedures for estimation and inference are discussed. Finally, the model proposed was illustrated using evasion’data from Statistics course of UFAM, available at Information System for Teaching from Universidade Federal do Amazonas-SIE/UFAM.