Explanation should include the benefits (e.g., increased outcome reliability) of combining many decision trees, each trained on a subset of data sampled at random, into a “forest” of decision trees.