Optimal Designs for the Evaluation of an Extremum Point
This paper studies the optimal experimental design for the evaluation of an extremum point of a quadratic regression function of one or several variables. Experimental designs which are locally optimal for arbitrary dimension k among all approximate designs are constructed (although for k > 1 an explicit form proves to be available only under a restriction on the location of the extremum point). The result obtained can be considered as an improvement of the last step of the well-known Box-Wilson procedure
KeywordsOptimal Design Extremum Point Regression Function Quadratic Regression Optimal Experimental Design
Unable to display preview. Download preview PDF.
- Karlin, S. and Studden, W. (1966). Tchebysheff Systems: with Application in Analysis and Statistics. New York: Wiley.Google Scholar
- Mandai, N.K. and Heiligers, B. (1992). Minimax designs for estimating the optimum point in a quadratic response surface. JSPI 31, 235–244.Google Scholar
- Mandai, N.K. (1978). On estimation of the maximal point of single factor quadratic response function. Bull. Calcutta Statist. Assoc. 27, 119–125.Google Scholar
- Müller, W.G. and Pötscher, B.M. (1992). Batch sequential design for a nonlinear estimation problem. In Model-Oriented Data Analysis 2 Eds V.V. Fedorov, W.G. Müller and I. Vuchkov, pp. 77–87. Heidelberg: Physica-Verlag.Google Scholar