Reliability Prediction with Inadequate Data

Although the accuracy of reliability prediction depends on both mathematical models and available data, the latter most often constitutes the limiting factor. Data inadequacies result in prediction uncertainties. If these uncertainties can be quantified, they need not negate the usefulness of predictions. In fact, quantified uncertainty may aid in system of configuration selection and in the allocation of further effort. The paper describes an approach utilizing information from diverse sources to predict reliability and estimate prediction uncertainty.