Fig. 2From: Boosting for high-dimensional two-class predictionResubstitution error rate of CART(5) (left panels) and stumps (CART(1), right panels). The figure reports mean resubstitution error rate of CART(5) and CART(1) for different number of variables and size of the training set (n train ). The true error rate in this setting is 0.5. The results are based on 1000 simulationsBack to article page