Optimal Designs for Estimation of Optimum Mixtures and Optimum Amount in a Multiresponse Mixture Experiment

Abstract

Optimal designs for estimating the parameters and also the optimum factor combinations in multi-response experiments have been considered by some authors. However, the existing literature on mixture experiments shows studies mainly in the single response case. In this paper an attempt has been made to investigate optimum designs for estimating optimum mixing proportions and also the optimum amount of mixture in a multi-response experiment. The pseudo-Bayesian approach has been used, and the support points of the optimum design are found to be the union of the support points of a weighted centroid design and a three-point symmetric design, with support points at the two extremes and one at the centre.