D-optimal design for ordinal responses in mixture experiments

Abstract

A mixture experiment is an experiment in which the response depends on the ingredients proportions. In organoleptic test, the response is ordinal. This research focused on how to generate the D-optimal design of ordinal response. The model belongs to Generalized Linear Model (GLM) with logit as the link function. To find the D-optimal design which is seeking the design that minimizing the covariance of parameter model, the point-exchange algorithm was constructed. There were three ingredients of a cookies recipe which were needed to mix meanwhile other ingredients were fixed. The D-optimal design of the three ingredients consisted of seven design points.