MODEL ERRORS: ANALYSIS BY CLASSIFICATION TREES AND IMPLICATIONS ON OPTIMAL DESIGN

Abstract eng:
In the search for the optimal design of a system, approximate models are often needed. Before such models are used, a regression is usually performed to fit the model to a more accurate, reference model. Inherent in such a fitting procedure is a statistical model for the errors. In this extended abstract, the model errors is examined for potential use in the optimal design problem. With the randomness in the errors, a probabilistic view is still needed, however, the approach developed herein is fundamentally different from that used in statistical regression. With the focus on design, it is shown that the outliers, which are ignored or otherwise discounted in statistical regression, can be used to identify system features that may potentially lead to improved design.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket