Recently, some interest has been focussed on the estimation of component life time distributions in coherent systems. For simplicity, most of the work assume independence between different component life times. Also, some specific dependence structure has been dealt with in some limited cases. In this work, we introduce a unified way of modelling the component life time distributions in a coherent system through the framework of multivariate failure times, which includes some arbitrary nature of dependence. Under independence, both parametric and nonparametric estimation of component life time distributions are discussed with continuous monitoring data and autopsy data as well. A specific dependence structure is also considered for this purpose.
Marginal Methods for Multivariate Failure Times Under Event-Dependent Censoring
