A note on using Bayes priors for Weibull distribution

Abstract

In this note, the practical use of priors for Bayes estimators, of the two parameters of the Weibull reliability model, is discussed in a technological context. The meaning of the priors as expression of virtual data samples is analyzed. The implications of physics of failures are also highlighted. The whole analysis shows a rational way to convert a really available technological knowledge into prior information, effectively and without a long elicitation process. A large Monte Carlo study, on both complete and censored samples of small size, shows the good properties of some estimators, which can exploit such a new approach.