Genetic Function Approximation Experimental Design (GFAXD): A New Method for Experimental Design
Thomas R. Kowar · Journal of Chemical Information and Computer Sciences · 1998
The application of Roger's Genetic Function Approximation (GFA) algorithm to experimental design data produces a population of model equations that contains the same regression equation as derived using Statistical Experimental Design analysis. GFA is able to identify this regression equation from fewer experiments than required by Statistical Experimental Design without the loss of information about higher order effects. A proposal for a novel method for the design, conduct, and analysis of experiments, Genetic Function Approximation Experimental Design (GFAXD), is presented. An example application of the GFAXD method to a theoretical 12-variable experimental design problem demonstrates the potential of this new method to significantly increase the productivity of designed experimentation.