A Complex Weapon System Bayes' Strategy Using Component Test Data 700625

The paper defines a “best” estimate of system effectiveness or reliability as one that minimizes the potential loss due to either overestimating or underestimating such a system figure of merit. A loss function is expressed in terms of the decision maker's order of preference for the consequences of either overestimation (underkill) or underestimation (overkill). However, in order for an estimate to be optimal, sufficient information at the system level should be provided. This lack or abundance of information is reflected in the variability of the measurement of system effectiveness or reliability. The variability and central tendency of this measurement are obtained from the variability and central tendency of the component data. This is achieved through the combined use of effectiveness or reliability models, Monte Carlo simulation and/or probability moments, and Bayesian statistics.