Strategic Data Collection for Optimal Use of Resources Based on Multiple Criteria

Christine Anderson-Cook, CCS-6 The LANL Statistical Sciences Group, CCS-6, has an established record of research in experiment design and resource allocation. The goal of these efforts is to provide a quantitatively justified approach for determining what future data should be collected based on specific programmatic goals. The Pareto front approach allows decision-makers to establish specific objectives on which to focus for future data collection and, to identify superior choices to consider further, and then provide graphical summaries to compare alternatives with the goal of identifying a best choice. The approach allows flexibility for how the different objectives are weighted in the optimization, while allowing examination of the robustness across different prioritizations. This article describes a simplified example of how the methods can be used to maximally reduce uncertainty associated with system and subsystem reliability estimates.