Curvature, Robustness, and Optimal Design in Applied Nonlinear Regression Modeling

• Activity Number: 486
• Type: Invited
• Date/Time: Thursday, August 10, 2006 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Timothy E. O'Brien*+
• Companies: Loyola University Chicago
• Keywords: experimental design ; lack of fit ; differential geometry ; optimality

Abstract

Researchers often find that nonlinear regression models are more applicable for modeling their processes than are linear ones. These researchers are thus often in a position of requiring optimal or near-optimal designs for a given nonlinear model. A common shortcoming of most optimal designs for nonlinear models used in practical settings, however, is that these designs typically focus on only (first-order) parameter variance or predicted variance and ignore the inherent nonlinearity of the assumed model function. Another shortcoming of optimal designs is that they often have only p support points, where p is the number of model parameters. This talk examines the reliability of Clarke's marginal curvature measures in practical settings and introduces a design criterion that combines variance minimization with nonlinearity minimization. Numerous illustrations will be provided.