Assessing Response’s Bias, Quality of Predictions, and Robustness in Multiresponse Problems

摘要: Optimization measures for evaluating compromise solutions in multiresponse problems formulated the Response Surface Methodology framework are proposed. The take into account desired properties of responses at optimal variable settings, namely, bias, quality predictions and robustness, which allow analyst to achieve interest feasible practice, namely case a method that does not consider objective function responses’ variance level correlation information is used. Two examples from literature show utility proposed measures.