1.Bootstrapping: It starts by randomly selecting a subset of the training data with replacement.