Bayesian Additive Regression Trees (BART) is a nonparametric ensemble method that models complex relationships by summing a collection of decision trees, each operating as a weak learner.