Optimal Factor Assignment for Supersaturated Designs

Abstract

This paper formulates factor assignment problem on supersaturated designs with considering all permutations of columns. This paper also gives an approximate algorithm that selects columns' as little correlation as possible. The algorithm is based on a random choice in order to prevent combinatorial explosion. As an example of application, the results for screening factors by supersaturated designs before and after the column permutation are compared in a simulation study, and the necessity of the permutation is also discussed.