A general construction method for mixed-level supersaturated design

  • Authors:

  • Shu Yamada;Michiyo Matsui;Tomomi Matsui;Dennis K. J. Lin;Takenori Takahashi

  • Affiliations:

  • Graduate School of Business Sciences, University of Tsukuba, Otsuka 3-29-1, Tokyo 112-0012, Japan;Department of Management Science, Tokyo University of Science, Kagurazaka, Tokyo 162-8601, Japan;Department of Mathematical Informatics, Graduate School of Information Science and Technology, The University of Tokyo, Hongo 7-3-1, Tokyo 113-8656, Japan;Department of Supply Chain and Information Systems, The Pennsylvania State University, University Park, PA 16802-3005, USA;Department of Management Science, Tokyo University of Science, Kagurazaka, Tokyo 162-8601, Japan

  • Venue:
  • Computational Statistics & Data Analysis
  • Year:
  • 2006

Quantified Score

Hi-index 0.03

Visualization

Abstract

When the number of the experimental variables is large, the first and most critical step is to identify the (few) active factors among those (many) candidate factors. Supersaturated design is shown to be helpful for such a critical first step. A general construction method for mixed-level supersaturated design is proposed. The newly constructed design has several advantages, including the flexibility for the number of runs and the assurance of upper bound of the (pairwise) dependency among all design columns. Specific applications to the construction of two-level and three-level mixed-level designs are discussed in detail.