Comparison of experimental designs combining process and mixture variables: Part II. Design evaluation on measured data

The construction of a small experimental design for a combination of process and mixture variables is a problem which has not been solved completely by now. In a previous paper we evaluated some designs with theoretical measures. This second paper evaluates the capabilities of the best of these designs in a practical application. For this purpose, data are obtained from a pharmaceutical problem according to these designs. The model selection on the basis of the small designs is compared with model selection on the basis of an extra data set. The prediction errors of the small designs are estimated using a test set.