Subjective Priors in Bayesian Estimation of Safety and Reliability Characteristics

The problem of estimation some unknown safety and reliability characteristics using nonparametric method of Bayesian estimation is considered in this paper. In many cases, the expert's opinion is the only way to obtain the priors. To construct the subjective priors we use some methods of elicitation and scoring. Construction of the nonparametric estimator for the uncensored data is based on Dirichlet process, which is key concept in the Ferguson theory. Two examples illustrate the presented concepts and theorems.