Home > Events > 2011 Seminars & Colloquia > BRADLEY JONES - JMP/SAS


Main Content

” D-optimal Design of Supersaturated Experiments with Suggestions for Analysis”
17 November 2011 from 4:00 PM to 5:00 PM
111 Tyson Bldg.
Add event to calendar

Supersaturated designs are experiments where the number factors exceeds the number of runs. As a result the model matrix is singular and ordinary least squares estimates of the parameters do not exist. However, if only a few of the factors have a pronounced effect on the response of interest, then it is possible to identify these factors and estimate their effects. Research in supersaturated design considers both how to construct the design and given the design how to analyze the data. In this talk I introduce a new optimality criterion for  evaluating supersaturated designs and provide some ideas for using this criterion to construct such designs. I will also connect this criterion to E(s^2) optimal designs and Bayesian D-optimal designs. The new designs have some special features that suggest a new analytical approach that is both simple and powerful.

Filed under: ,