(Received March 27, 1996; revised April 23, 1997)
Abstract. In this paper we introduce a finite population version of the mean residual life-time (MRL) function and the hazard function, and study Bayesian estimation of these functions. The unknown parameter is the complete set (y1, ...., yN) of lifetimes of the N units which constitute the complete population. A hierarchical type prior is used, where the yi's are assumed conditionally independent given a random parameter theta. The data consists of a random sample of n values of yi. The Bayes estimators of MRL and hazard functions, respectively, are then obtained as the posterior expectations of the unknown functions.
Key words and phrases: Finite population, mean residual life-time function, hazard function, exponential distribution, prior distribution, posterior distribution, exchangeability, type I censoring, type II censoring.