Bayesian Analysis for Dependent Progressively Censored Weibull Competing Risks Using Copulas

Abstract

In many reliability studies, the experimental units may fail due to one of several causes of failure. It is usually assumed that the competing risks of failure are independent. In many practical situations, however, the interpretation of the failure modes makes the assumption of independence unreasonable. Copulas are considered an effective tool for modeling the dependence structure among the multiple competing risks. This paper presents Bayesian analysis of progressively Type-II censored dependent competing risks data using copulas. The analysis is performed under the assumption of binomial progressive random removals and Weibull failure times, where unit failure occurs due to only one of the competing risks. Bayesian point and interval estimates of the unknown parameters are derived using different Archimedean copulas with non-conjugate prior distributions. A simulation study is carried out to assess the performance of the proposed techniques under different dependence structures. A real data set is analyzed for illustrative purposes.