Editing Random forest (section)

===ExtraTrees===
Adding one further step of randomization yields ''extremely randomized trees'', or ExtraTrees. As with ordinary random forests, they are an ensemble of individual trees, but there are two main differences: (1) each tree is trained using the whole learning sample (rather than a bootstrap sample), and (2) the top-down splitting is randomized: for each feature under consideration, a number of ''random'' cut-points are selected, instead of computing the locally ''optimal'' cut-point (based on, e.g., [[information gain]] or the [[Gini impurity]]). The values are chosen from a uniform distribution within the feature's empirical range (in the tree's training set). Then, of all the randomly chosen splits, the split that yields the highest score is chosen to split the node.

Similar to ordinary random forests, the number of randomly selected features to be considered at each node can be specified. Default values for this parameter are <math>\sqrt{p}</math> for classification and <math>p</math> for regression, where <math>p</math> is the number of features in the model.<ref>{{Cite journal | doi = 10.1007/s10994-006-6226-1| title = Extremely randomized trees| journal = Machine Learning| volume = 63| pages = 3–42| year = 2006| vauthors = Geurts P, Ernst D, Wehenkel L | url = http://orbi.ulg.ac.be/bitstream/2268/9357/1/geurts-mlj-advance.pdf| doi-access = free}}</ref>