Online Program Home
My Program

Abstract Details

Activity Number: 34 - Linear Models for Large or Complex Data
Type: Contributed
Date/Time: Sunday, July 29, 2018 : 2:00 PM to 3:50 PM
Sponsor: Section on Physical and Engineering Sciences
Abstract #329403 Presentation
Title: Parameter Subset Selection for Mixed-Effects Models
Author(s): Kathleen Schmidt* and Ralph C. Smith and Jason Bernstein and Ana Kupresanin
Companies: Lawrence Livermore National Laboratory and North Carolina State University and Lawrence Livermore National Laboratory and Lawrence Livermore National Laboratory
Keywords: Sensitivity Analysis; Mixed-effects; Model Selection; Uncertainty Quantification; Materials Science
Abstract:

Mixed-effects are often used to model individual variation within a population; however, uncertainty quantification for such models remains an open area of research. Sensitivity analysis proves particularly challenging since traditional methods, such as Morris screening, are generally ineffective for mixed-effects models. As a result, the current literature focuses on model selection, by which insensitive parameters are fixed or removed from the model, but such techniques are limited in that they are computationally prohibitive for large problems due to the number of possible models that must be tested. We introduce a parameter subset selection (PSS) method, based on standard error estimates, for mixed-effects models. This PSS method can be used to limit the number of possible models to be tested and ease the computational strain of model selection. We compare our method to existing techniques and use our PSS method to select the significant parameters of a mixed-effects materials strength model. LLNL-ABS-745166. This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program