Statistical Analysis of Software Failure Data

Abstract

In Chapters 2 and 3, we introduced several models for describing our uncertainties about the software failure process. These models involved unknown parameters, often denoted by Greek symbols. The parameters entered into the picture because of our invoking the law of the extension of conversation, as a way of simplifying the probability specification process. The parameters being unobservable, our uncertainty about them was described by a prior distribution. The prior distribution is specific to an individual and may vary from individual to individual.