When experimentation is expensive and the number of factors is large, supersaturated designs can be helpful. They are essentially fractional factorial designs in which the number of factors is greater than the number of experimental runs. Previous studies have focused on two-level supersaturated designs. This paper presents a new class of three-level supersaturated designs with an equal occurrence property. It is shown that designs generated by such a universal construction method result in a low dependency among all columns.
All Science Journal Classification (ASJC) codes
- Statistics and Probability
- Statistics, Probability and Uncertainty