Bayesian statistics in reliability engineering

[Tarik]

Hi,
I would like to start a discussion here about the applications of Bayesian statistics in reliability engineering. Do you have examples, stories, applications or problems to share with us? Where have you find Bayesian statistics to be useful in your applications?

[Nath]

Wei-bayes or classic bayesian with priors?

[tarik]

Some people refer to the 1 parameter Weibull distribution (you specify beta and solve for eta) as Wei-bayes which is not a true Bayesian Weibull distribution. If you are actually using a prior distribution, then that becomes a true Wei-Bayes.

[Nath]

Its a systematic 6 steps method, learnt at U of Arizona. Tried to use it in the real world of normal priors and thus the posterior is also normal. The bayes estimates were quite close to the classical, and the bounds were narrower than the classical.

[Steve]

I recently attempted to convince some in my company that it would be worthwhile to find the distributions for the betas associated with our warranty data on parts. From this we could create a "beta distribution library" for systems and components. It seemed like a idea worth pursuing, even though our data is not perfect by any means. Has anyone else tried doing this? Or do you only use betas from testing or other "controlled" approaches?

[HaroldS]

Steve,

I see that as a worthwhile pursuit. If you have the data there is no reason why you should not do that in my humble opinion.