Extra Tree Classifier#

An Extremely Randomized Classification Tree that splits the training set at a random point with the lowest entropy among m features. Extra Trees are useful in ensembles such as Random Forest or AdaBoost as the weak classifier or they can be used on their own. The strength of Extra Trees are their computational efficiency and increased variance of predictions.

Interfaces: Estimator, Learner, Probabilistic, Persistable

Data Type Compatibility: Categorical, Continuous


# Param Default Type Description
1 max depth PHP_INT_MAX int The maximum depth of a branch in the tree.
2 max leaf size 3 int The max number of samples that a leaf node can contain.
3 max features Auto int The max number of features columns to consider when determining a best split.
4 min purity increase 1e-7 float The minimum increase in purity necessary for a node not to be post pruned.

Additional Methods#

Return the normalized feature importances i.e. the proportion that each feature contributes to the overall model, indexed by feature column:

public featureImportances() : array

Display a human readable text representation of the decision tree:

public printrules() : void

Return the height of the tree:

public height() : int

Return the balance factor of the tree:

public balance() : int


use Rubix\ML\Classifiers\ExtraTreeClassifier;

$estimator = new ExtraTreeClassifier(50, 3, 4, 1e-7);


  • P. Geurts et al. (2005). Extremely Randomized Trees.