Bayesian D-Optimal Design of Experiments with Quantitative and Qualitative Responses

Systems with both quantitative and qualitative responses are widely encountered in many applications. Design of experiment methods needed when experiments conducted to study such systems. Classic experimental design unsuitable here because they often focus on one type response. In this paper, we develop a Bayesian D-optimal method for continuous binary Both noninformative conjugate informative prior distributions the unknown parameters considered. The proposed criterion has meaningful interpretations regarding D-optimality models types responses. An efficient point-exchange search algorithm is developed construct local designs given parameter values. Global obtained by accumulating frequencies points designs, where sampled from distributions. performances evaluated through two examples.