Using Maple to Design a Complex Mixture Experiment

Abstract

The design for an experiment can sometimes be taken directly from one of many books on the subject. Often, the design must be at least slightly modified. All too often, a complex experiment requires a design that cannot be found ready-made. Most experimental design software has focused on 2- and 3-level factorial designs and other specialized designs (e.g. Box-Benkin). Adapting or creating an experimental design can be a daunting task for many experimenters and even applied statisticians. Little experimental design software exists to tackle the more complex design problems. Maple can be used to produce complex designs and by using its built-in procedures the programming task is greatly simplified. In the following, Maple is used to design a complex mixture experiment consisting of seven factors.