Optimized CART and Extra Tree memory and node splitting

CHANGELOG.md

@@ -41,6 +41,7 @@
     - Optimized CART binary categorical node splitting
     - Interval Discretizer outputs numeric string categories
     - Renamed Random Hot Deck Imputer
+    - Changed order of decision tree hyper-parameters
 
 - 0.4.1
     - Optimized CART node splitting for low variance continuous features

docs/classifiers/classification-tree.md

@@ -12,14 +12,14 @@ A binary tree-based learner that greedily constructs a decision map for classifi
 |---|---|---|---|---|
 | 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 | maxFeatures | Auto | int | The max number of feature columns to consider when determining a best split. |
-| 4 | minPurityIncrease | 1e-7 | float | The minimum increase in purity necessary for a node *not* to be post pruned during tree growth. |
+| 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. |
 
 ## Example
 ```php
 use Rubix\ML\Classifiers\ClassificationTree;
 
-$estimator = new ClassificationTree(10, 7, 4, 0.01);
+$estimator = new ClassificationTree(10, 5, 0.001, null);
 ```
 
 ## Additional Methods

docs/classifiers/extra-tree-classifier.md

@@ -12,14 +12,14 @@ An *Extremely Randomized* Classification Tree that recursively chooses node spli
 |---|---|---|---|---|
 | 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 | maxFeatures | Auto | int | The max number of feature columns to consider when determining a best split. |
-| 4 | minPurityIncrease | 1e-7 | float | The minimum increase in purity necessary for a node *not* to be post pruned during tree growth. |
+| 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. |
 
 ## Example
 ```php
 use Rubix\ML\Classifiers\ExtraTreeClassifier;
 
-$estimator = new ExtraTreeClassifier(50, 3, 4, 1e-7);
+$estimator = new ExtraTreeClassifier(50, 3, 1e-7, 10);
 ```
 
 ## Additional Methods

docs/regressors/extra-tree-regressor.md

@@ -12,14 +12,14 @@
 |---|---|---|---|---|
 | 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 | maxFeatures | Auto | int | The max number of feature columns to consider when determining a best split. |
-| 4 | minPurityIncrease | 1e-7 | float | The minimum increase in purity necessary for a node *not* to be post pruned during tree growth. |
+| 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. |
 
 ## Example
 ```php
 use Rubix\ML\Regressors\ExtraTreeRegressor;
 
-$estimator = new ExtraTreeRegressor(30, 3, 20, 0.05);
+$estimator = new ExtraTreeRegressor(30, 5, 0.05, null);
 ```
 
 ## Additional Methods

docs/regressors/regression-tree.md

@@ -12,14 +12,14 @@ A decision tree based on the CART (*Classification and Regression Tree*) learnin
 |---|---|---|---|---|
 | 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 | maxFeatures | Auto | int | The max number of feature columns to consider when determining a best split. |
-| 4 | minPurityIncrease | 1e-7 | float | The minimum increase in purity necessary for a node *not* to be post pruned during tree growth. |
+| 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. |
 
 ## Example
 ```php
 use Rubix\ML\Regressors\RegressionTree;
 
-$estimator = new RegressionTree(20, 2, null, 1e-3);
+$estimator = new RegressionTree(20, 2, 1e-3, 10);
 ```
 
 ## Additional Methods

src/Classifiers/ClassificationTree.php

@@ -24,6 +24,12 @@
 use Rubix\ML\Exceptions\RuntimeException;
 
 use function Rubix\ML\argmax;
+use function count;
+use function array_fill;
+use function array_combine;
+use function array_replace;
+use function array_count_values;
+use function array_map;
 
 /**
  * Classification Tree
@@ -46,9 +52,9 @@ class ClassificationTree extends CART implements Estimator, Learner, Probabilist
     use AutotrackRevisions;
 
     /**
-     * The zero vector for the possible class outcomes.
+     * The list of possible class outcomes.
      *
-     * @var float[]
+     * @var list<string>
      */
     protected $classes = [
         //
@@ -57,16 +63,16 @@ class ClassificationTree extends CART implements Estimator, Learner, Probabilist
     /**
      * @param int $maxHeight
      * @param int $maxLeafSize
-     * @param int|null $maxFeatures
      * @param float $minPurityIncrease
+     * @param int|null $maxFeatures
      */
     public function __construct(
         int $maxHeight = PHP_INT_MAX,
         int $maxLeafSize = 3,
-        ?int $maxFeatures = null,
-        float $minPurityIncrease = 1e-7
+        float $minPurityIncrease = 1e-7,
+        ?int $maxFeatures = null
     ) {
-        parent::__construct($maxHeight, $maxLeafSize, $maxFeatures, $minPurityIncrease);
+        parent::__construct($maxHeight, $maxLeafSize, $minPurityIncrease, $maxFeatures);
     }
 
     /**
@@ -108,8 +114,8 @@ public function params() : array
         return [
             'max height' => $this->maxHeight,
             'max leaf size' => $this->maxLeafSize,
-            'max features' => $this->maxFeatures,
             'min purity increase' => $this->minPurityIncrease,
+            'max features' => $this->maxFeatures,
         ];
     }
 
@@ -137,7 +143,7 @@ public function train(Dataset $dataset) : void
             new LabelsAreCompatibleWithLearner($dataset, $this),
         ])->check();
 
-        $this->classes = array_fill_keys($dataset->possibleOutcomes(), 0.0);
+        $this->classes = $dataset->possibleOutcomes();
 
         $this->grow($dataset);
     }
@@ -207,7 +213,9 @@ public function probaSample(array $sample) : array
         /** @var \Rubix\ML\Graph\Nodes\Best $node */
         $node = $this->search($sample);
 
-        return array_replace($this->classes, $node->probabilities()) ?? [];
+        $template = array_combine($this->classes, array_fill(0, count($this->classes), 0.0)) ?: [];
+
+        return array_replace($template, $node->probabilities());
     }
 
     /**
@@ -237,20 +245,20 @@ protected function terminate(Labeled $dataset) : Best
     }
 
     /**
-     * Compute the gini impurity of a labeled dataset.
+     * Calculate the impurity of a set of labels.
      *
-     * @param \Rubix\ML\Datasets\Labeled $dataset
+     * @param list<string|int> $labels
      * @return float
      */
-    protected function impurity(Labeled $dataset) : float
+    protected function impurity(array $labels) : float
     {
-        $n = $dataset->numRows();
+        $n = count($labels);
 
         if ($n <= 1) {
             return 0.0;
         }
 
-        $counts = array_count_values($dataset->labels());
+        $counts = array_count_values($labels);
 
         $gini = 0.0;
 

src/Classifiers/ExtraTreeClassifier.php

@@ -24,6 +24,12 @@
 use Rubix\ML\Exceptions\RuntimeException;
 
 use function Rubix\ML\argmax;
+use function count;
+use function array_fill;
+use function array_combine;
+use function array_replace;
+use function array_count_values;
+use function array_map;
 
 /**
  * Extra Tree Classifier
@@ -47,9 +53,9 @@ class ExtraTreeClassifier extends ExtraTree implements Estimator, Learner, Proba
     use AutotrackRevisions;
 
     /**
-     * The zero vector for the possible class outcomes.
+     * The list of possible class outcomes.
      *
-     * @var float[]
+     * @var string[]
      */
     protected $classes = [
         //
@@ -58,16 +64,16 @@ class ExtraTreeClassifier extends ExtraTree implements Estimator, Learner, Proba
     /**
      * @param int $maxHeight
      * @param int $maxLeafSize
-     * @param int|null $maxFeatures
      * @param float $minPurityIncrease
+     * @param int|null $maxFeatures
      */
     public function __construct(
         int $maxHeight = PHP_INT_MAX,
         int $maxLeafSize = 3,
-        ?int $maxFeatures = null,
-        float $minPurityIncrease = 1e-7
+        float $minPurityIncrease = 1e-7,
+        ?int $maxFeatures = null
     ) {
-        parent::__construct($maxHeight, $maxLeafSize, $maxFeatures, $minPurityIncrease);
+        parent::__construct($maxHeight, $maxLeafSize, $minPurityIncrease, $maxFeatures);
     }
 
     /**
@@ -138,7 +144,7 @@ public function train(Dataset $dataset) : void
             new LabelsAreCompatibleWithLearner($dataset, $this),
         ])->check();
 
-        $this->classes = array_fill_keys($dataset->possibleOutcomes(), 0.0);
+        $this->classes = $dataset->possibleOutcomes();
 
         $this->grow($dataset);
     }
@@ -208,7 +214,9 @@ public function probaSample(array $sample) : array
         /** @var \Rubix\ML\Graph\Nodes\Best $node */
         $node = $this->search($sample);
 
-        return array_replace($this->classes, $node->probabilities()) ?? [];
+        $template = array_combine($this->classes, array_fill(0, count($this->classes), 0.0)) ?: [];
+
+        return array_replace($template, $node->probabilities());
     }
 
     /**
@@ -240,20 +248,20 @@ protected function terminate(Labeled $dataset) : Best
     }
 
     /**
-     * Compute the entropy of a labeled dataset.
+     * Calculate the impurity of a set of labels.
      *
-     * @param \Rubix\ML\Datasets\Labeled $dataset
+     * @param list<string|int> $labels
      * @return float
      */
-    protected function impurity(Labeled $dataset) : float
+    protected function impurity(array $labels) : float
     {
-        $n = $dataset->numRows();
+        $n = count($labels);
 
         if ($n <= 1) {
             return 0.0;
         }
 
-        $counts = array_count_values($dataset->labels());
+        $counts = array_count_values($labels);
 
         $entropy = 0.0;