BAYESIAN NONPARAMETRIC ESTIMATION OF A MONOTONE HAZARD RATE

Abstract

The Bayesian nonparametric estimation of a monotone hazard rate for a reliability model is considered. In general a posterior mean of a mixture hazard rate with respect to a gamma-type process prior could be written as an average over all possible partitions of the first K integers, where K is the number of the complete data. Numerical approximations to posterior quantities based on Gibbs samplers which have stationary distributions on partitions are studied. Examples are given.