Skip to content


Extra Tree Regressor#

Extremely Randomized Regression Trees differ from standard Regression Trees in that they choose candidate splits at random rather than searching the entire feature column for the best value to split on. Extra Trees are also faster to build and their predictions have higher variance than a regular decision tree regressor.

Interfaces: Estimator, Learner, Ranks Features, Persistable

Data Type Compatibility: Categorical, Continuous


# Name Default Type Description
1 maxHeight PHP_INT_MAX int The maximum height of the tree.
2 maxLeafSize 3 int The max number of samples that a leaf node can contain.
3 minPurityIncrease 1e-7 float The minimum increase in purity necessary to continue splitting a subtree.
4 maxFeatures Auto int The max number of feature columns to consider when determining a best split.


use Rubix\ML\Regressors\ExtraTreeRegressor;

$estimator = new ExtraTreeRegressor(30, 5, 0.05, null);

Additional Methods#

Export a Graphviz "dot" encoding of the decision tree structure.

public exportGraphviz() : Encoding

Return the number of levels in the tree.

public height() : ?int

Return a factor that quantifies the skewness of the distribution of nodes in the tree.

public balance() : ?int


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