A1204
Title: Model-robust process optimization for data with a multistratum structure
Authors: Po Yang - University of Manitoba (Canada) [presenting]
Abstract: Response optimization is a process of identifying the input variable settings that optimize the response. Multistratum design arises naturally in industrial experiments due to the inconvenience and impracticality of complete randomization. Accounting for model uncertainty, Bayesian model averaging and the predictive approach are applied to investigate the optimization problem for data with a multistratum structure. Bridge sampling is used to estimate marginal likelihoods and hence posterior model probabilities. The goal of the optimization is to identify factor settings that maximize the probability that the response falls within a given range. The method is illustrated with two examples.
