Modelos de regressão log-Birnbaum-Saunders generalizados para dados com censura intervalar.

Abstract

In this work, we propose the model of regression log-Birnbaum-Saunders generalized to analyze data with interval censored. The score functions and the observed Fisher information matrix were obtained, as well as the process for estimating of the parameters was discussed. As measure of influence, we considered the likelihood displacement under several schemes of perturbation. The normal curvatures of local influence were derived and we conducted a residual analysis based on residuals Cox-Snell adjusted, Martingale and modified deviance. A Monte Carlo simulation was carried in order to investigate the behavior of empiric distribution of the proposed residuals. Finally, an application with real data is presented.