keywords design of experiment, supersaturated split-plot designs, bayesian optimality, shrinkage; regression