Análise de resíduos em modelos de tempo de vida acelerado Log-Birnbaum-Saunders bivariado.

Abstract

In this work, we propose a bivariate log-Birnbaum-Saunders accelerated lifetime model and present some residuals to evaluate the fit this model. In this sense, we investigated the capacity of the proposed residues to detect bad specifications in the stochastic and systematic components of the model, as well as to detect the presence of outliers. We will do this through a graphical analysis and through a measure of quality of fit. In addition, we present a version for censored data on the right and we analyze through simulation studies if such adjusted residuals show an improvement in comparison with the unadjusted residues, in terms of evaluating the fit of the model in case we are dealing with censored data on the right. Finally, we illustrate the methodology presented through a set of real data.