CMStatistics 2016: Start Registration
View Submission - CMStatistics
B1075
Title: Some relations between optimum on-the-average designs and maximin designs Authors:  Hans Nyquist - Stockholm University (Sweden) [presenting]
Abstract: Construction of optimal experimental designs often requires knowledge of unknown parameter values. Two proposed approaches out of this dilemma are using optimum on-the-average designs, which use a weighted average of criterion functions, and maximin designs, which find the best guaranteed design as the parameters vary in a specified subset of the parameter space. Some relations between these two approaches are discussed and utilized to construct an algorithm for constructing maximin designs. Specifically, examples of maximin efficient designs are discussed.