Return the predicted outcome for each treatment made by a policy learner for each point in the features