Extremely Randomized TreesExtremely randomized trees are another tree-based ensemble method for classificationproblems. Geurts et al. [45] state that the cut-point, determined in the node ofa decision tree, is associated with high variance in tree-based models, such as CARTand C4.5. It is therefore responsible for a significant part of the error rates of tree-basedmethods [45]. Rather than relying on discriminative thresholds in the decisionnode, the split is randomly selected in this algorithm to mitigate these problems.