Estimadores não paramétricos para dados com censura

Abstract

In this paper we study the reliability of systems with connected components in series and parallel. For systems in series, the device fails when the first component fails. Although in the parallel systems this happens when the last component fails. We define the distribution functions, sub-distribution functions and their respective properties. Here we present the Bayesian nonparametric estimator for the components of both systems, illustrating by examples. For both cases we performed a comparative study between the Bayesian nonparametric estimator and the Kaplan-Meier method.