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
Parameters#
| # | Param | Default | Type | Description |
|---|---|---|---|---|
| 1 | max height | PHP_INT_MAX | int | The maximum height of 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 feature 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 during tree growth. |
Example#
use Rubix\ML\Regressors\ExtraTreeRegressor;
$estimator = new ExtraTreeRegressor(30, 3, 20, 0.05);
Additional Methods#
Return a human-readable text representation of the decision tree ruleset:
public rules(?array $header = null) : string
Return the height of the tree i.e. the number of layers:
public height() : int
Return the balance factor of the tree:
public balance() : int
References#
- P. Geurts et al. (2005). Extremely Randomized Trees.