Abstract
As a means of increasing the quality and productivity of R&D efforts, we provide an integrated collection of experimental design tools for use by researchers at the Becton Dickinson Research Center. The bases for these tools are a set of well-accepted guidelines for experimental procedures and a tabulation of practical experimental designs. The designs themselves are provided to researchers in an easy-to-use personal computer (PC) spreadsheet format. The data can then be easily exported to a commercial PC statistics package. A PC based expert system, called Dexter, assists researchers in following the guidelines for selecting which of the tabulated designs should be used for a given experiment.
Dexter incorporates the expertise of a consulting statistician in helping a user select a design from among 35 designs for industrial screening experiments. The designs considered include from three to eleven experimental factors with a maximum sample size of 32 runs. The evaluation is based on the calculation of design scores which model guidelines for selecting experimental designs. A graph isomorphism algorithm is used to match particular design characteristics such as estimable two-factor interactions. Smart search techniques and precomputing of graph representations minimize run times and make the PC platform practical. The window-oriented, menu-driven interface is intuitive and easy-to use.
Similar content being viewed by others
Explore related subjects
Discover the latest articles, news and stories from top researchers in related subjects.References
P.D. Haaland,Experimental Design in Biotechnology (Marcel Dekker, 1989).
G.J. Hahn, Experimental design in a complex world, Technometrics 26 (1984) 19–31.
G.E.P. Box, W.G. Hunter and J.S. Hunter,Statistics for Experimenters (Wiley, 1978).
R.F. Liddle and P.D. Haaland, Efficient nonstandard screening designs, presented at the Annual Meetings of the American Statistical Association, New Orleans, LA (1988).
C.J. Nachtsheim, Tools for computer aided design of experiments, J. Quality Technol. 19 (1987) 132–160.
P.D. Haaland, D. Yen and R.F. Liddle, An expert system for the design of screening experiments,Proc. Statistical Computing Section of the American Statistical Association (1985) pp. 223–228.
P.D. Haaland, D. Yen and R.F. Liddle, An expert system for experimental design,Proc. Statistical Computing Section of the American Statistical Association (1986) pp. 78–87.
P.D. Haaland, J.C. Lusth, R.F. Liddle and D.S. Wilson, Dexter: An expert system for evaluating experimental designs,Proc. Statistical Computing Section of the American Statistical Association (1988) pp. 63–72.
I. Bruha, Recognition of waveforms: Neural net approach,Proc. University of Maine Spring Symp. on Artificial Intelligence and Intelligent Tutoring Systems (1989) pp. 101–137.
G. Taguchi,Introduction to Quality Engineering (Asian Productivity Organization and American Supplier Institute, Dearborn, Michigan, 1986).
S.A. Cook, The complexity of theorem proving procedures,Proc. 3rd ACM Symp. on Theory of Computing (1971) pp. 151–158.
K. Melhorn,Graph Algorithms and NP-Completeness (Springer, 1984).
K.-L. Tsui and A. Shoemaker, An automatic experiment planner for robust design, presented at the 33rd Annual Fall Technical Conf., Houston, TX (1989).
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Haaland, P.D., Lusth, J.C., Liddle, R.F. et al. Dexter: A guide to selecting the best design for an industrial screening experiment. Ann Math Artif Intell 2, 179–195 (1990). https://doi.org/10.1007/BF01531005
Issue Date:
DOI: https://doi.org/10.1007/BF01531005