A group orthogonal supersaturated design (GOSSD) is constructed to facilitate the identification of active effects. These designs group factors such that within a group, factors are correlated but between groups, factors are orthogonal. This structure leads to two analysis features:
• Fake, or dummy, factors assigned to the first group are used to estimate variance, assuming that the second- or higher-order effects are negligible.
• Between-group orthogonality enables a two-step model selection procedure.
The Fit Group Orthogonal Supersaturated Platform performs a two-step model selection procedure, enables you to view a regression plot of all factors, and then provides a model for the active effects with the option to examine this model with the Fit Model platform.