Simulating Responses and Fitting Definitive Screening Designs
Sep 7, 2017 1:15 PM
Jones DSD St Louis 2017.jrn
Bradley Jones, PhD, JMP Principal Research Fellow, SAS
Joseph Morgan, JMP Principal Research Statistician Developer, SAS Ryan Lekivetz, PhD, JMP Senior Research Statistician Developer, SAS
Definitive screening designs (DSDs) are a new class of orthogonal screening designs that provide several desirable features. Main effects are uncorrelated with each other and also uncorrelated with any second-order effect (i.e., two-factor interactions and pure quadratic effects). Also, second-order effects may be correlated but are never confounded with each other. For DSDs with six or more factors, any three-factor projection is capable of fitting the full quadratic model with high efficiency. JMP 13 has introduced a new tool for fitting DSDs, which takes explicit advantage of their special structure. JMP 13 also provides a new and powerful tool for simulating responses for DSDs. This presentation will demonstrate the utility of these two new tools using practical examples.