On Prediction of Survival Time for Individual Systems

The problem of predicting length of life (survival time) of an individual system is discussed. The predictions are based on previous test data on the types of components involved in the system. As a preliminary the classical, Bayesian, and fiducial probability approaches to prediction are outlined. The use of prediction techniques is then discussed by means of examples, with some special attention being given to the problem of predicting the life of an r-out-of-k:G system whose components have exponentially distributed lifetimes.