Strategies for Supersaturated Screening Experiments

The first step in experimentation is typically a screening experiment, with the goal of finding the important few factors out of the many. In some cases, the number of potential factors is large and the number of experiments is limited by temporal or economic constraints – leading one to consider using a supersaturated design (SSD) where potential factors outnumber experimental runs.

Despite the vast amount of literature on SSDs, there is scant record of their use in practice. Maria Weese, an Associate Professor at Miami University, contends this imbalance is due to conflicting recommendations regarding SSD use in the literature, as well as the designs’ inability to meet practitioners’ analysis expectations. To address these issues, she first summarizes practitioner concerns and expectations of SSDs when pairing a design construction method with a particular analysis method. She explains how the choice of pairing is dependent on the screening objective.

Best suited for those interested in:

• How supersaturated designs could be beneficial to your screening experiments.
• How and when to consider using supersaturated designs.
• How to choose from various types of supersaturated design and analysis pairings.

Meet the speaker

Maria Weese
Farmer School of Business, Miami University

Maria Weese is an Associate Professor of Information Systems and Analytics at the Farmer School of Business, Miami University. Her research focuses on experimental design, machine learning applications in process monitoring, and statistical methodologies. She holds a Ph.D. and a master's degree in statistics from the University of Tennessee and earned her undergraduate degree in chemical engineering from Virginia Tech.

Before pursuing graduate studies, she worked as a process engineer in southwest Virginia. Weese serves as an associate editor for Technometrics and is a recipient of the Lloyd S. Nelson Award.