Abstract
Constructing the Bayesian optimal design depends on the choice of a prior distribution for the unknown parameter. Lacking informative or historical knowledge of the parameter, a parametric Bayesian approach cannot be expected in complex statistical problems. In this regard, a nonparametric Bayesian approach can be used, in which random prior distribution is considered. The Dirichlet process is employed as a prior on the space of distribution functions. In this paper, a non-parametric Bayesian approach is incorporated into an optimal design criterion. This method is illustrated by an example.