Inference for Weibull competing risks model with partially observed failure causes under generalized progressive hybrid censoring (Q2292017)

The authors consider a competing risks model in which failures occur at the minimum of two Weibull-distributed times (with common shape parameter \(\alpha\) and differing scale parameters \(\beta_1\) and \(\beta_2\)). A generalized progressive hybrid censoring scheme is used, which aims to sensibly limit the duration of the experiment based on the number of failures observed. Partially observed data is allowed, in the sense that the cause of component failure may be unknown for each observed failure time, and the experiment may end before all failures are observed. The authors derive point and interval estimates of the Weibull parameters in this setting, in both classical and Bayesian frameworks. They consider both the ordered case (where it is known that \(\beta_1<\beta_2\)) and the unordered case (without any such restriction on the underlying parameters). Results are illustrated with some numerical examples.