ENBIS Spring Meeting 2018

4 – 6 June 2018; Florence, Italy Abstract submission: 17 November 2017 – 20 April 2018

Bayesian Factor Discrimination in Definitive Screening Designs

5 June 2018, 15:25 – 15:50


Submitted by
Victor Aguirre
Victor Aguirre (Statistics Department, ITAM), Ernesto Barrios (Statistics Department, ITAM), Sofia Huerta (Universidad Veracruzana)
Definitive Screening Designs are a class of experimental designs that under factor sparsity allow the possibility of estimating main, interaction and quadratic effects. Then a factor could influence the response due to main or interaction or quadratic effects. Using the approach presented in this work we allow for this possibility by evaluating the evidence shown by the posterior probabilities of a set of the most relevant models where the factor is present. This approach contrasts with strategies where only the evidence of individual effects is considered. The approach is calibrated to have an experiment wise error rate of ten percent. We show with examples that is gives much better results than forward stepwise procedures usually present in commercial statistical packages. The procedure is programmed in an R based graphical interphase.

Return to programme