38 Pages Posted: 7 Jan 2016 Last revised: 2 Apr 2017
Date Written: April 1, 2017
The identification of the optimal design of products or packaging is challenged when attributes and their levels interact. Firms recognize this by testing product profiles prior to launch, where the effects of interactions can only be revealed in the head-to-head comparison of a small number of finalists. A difficulty in conducting analysis for product design is dealing with the high dimensionality of the design space, which also leads to the selection of promising finalists for testing to be challenging. We propose an experimental criterion of the expected improvement in market share for efficiently testing product profiles with high potential in sequential experiments. Then, we incorporate a stochastic search variable selection method to selectively estimate relevant interactions among the attributes. A validation experiment confirms that our proposed method leads to improved design concepts in a high-dimensional space compared to alternative methods.
Keywords: Design Criteria, Expected Improvement, Interaction Effects, Stochastic Search Variable Selection
Suggested Citation: Suggested Citation
Joo, Mingyu and Thompson, Michael L and Allenby, Greg M., Optimal Product Design by Sequential Experiments in High Dimensions (April 1, 2017). Available at SSRN: https://ssrn.com/abstract=2711333 or http://dx.doi.org/10.2139/ssrn.2711333