Reliability estimation based on general progressive censored data from the Weibull model: comparison between Bayesian and classical approaches

Abstract

Inthis article, we consider the problem of estimating the parameters and reliability function of the two-parameter Weibull model on the basis of a progressive Type-II censored sample. We consider both classical and Bayesian approaches. In the Bayesian framework, we suggest a bivariate prior density for the two unknown parameters. Assuming the squared error loss function, we derive exact forms of the Bayes estimates. Further, we consider non-informative priors. To assess the accuracy of the resulting estimates, we conduct simulation experiments. In such experiments, we calculate the estimated risks (ER’s) and mean squared errors (MSE’s) of the Bayes estimates and compare them with the corresponding mean squared errors (MSE’s) of the maximum likelihood estimates. In addition, we calculate the relative efficiency between the considered estimates. Finally, we draw some concluding remarks.