lamp.extratrees

Type members

Classlikes

case class ClassificationLeaf(targetDistribution: Seq[Double]) extends ClassificationTree
Companion:
object
Companion:
class
case class ClassificationNonLeaf(left: ClassificationTree, right: ClassificationTree, splitFeature: Int, cutpoint: Double, splitMissingIsLess: Boolean) extends ClassificationTree
Companion:
object
sealed trait ClassificationTree
Companion:
object
Companion:
class
case class RegressionLeaf(targetMean: Double) extends RegressionTree
Companion:
object
Companion:
class
case class RegressionNonLeaf(left: RegressionTree, right: RegressionTree, splitFeature: Int, cutpoint: Double, splitMissingIsLess: Boolean) extends RegressionTree
Companion:
object
Companion:
class
sealed trait RegressionTree
Companion:
object
Companion:
class

Value members

Concrete methods

def buildForestClassification(data: Mat[Double], target: Vec[Int], sampleWeights: Option[Vec[Double]], numClasses: Int, nMin: Int, k: Int, m: Int, parallelism: Int, seed: Long): Seq[ClassificationTree]

Train an extratrees classifier forest

Train an extratrees classifier forest

Value parameters:
k

number of features to consider in each split step. The best among these will be chosen.

m

number of trees

nMin

minimum sample size for splitting a node

seed

Returns a list of ClassificationTree objects which can be passed to predictClassification

def buildForestRegression(data: Mat[Double], target: Vec[Double], nMin: Int, k: Int, m: Int, parallelism: Int, seed: Long): Seq[RegressionTree]

Train an extratrees regression forest

Train an extratrees regression forest

Value parameters:
k

number of features to consider in each split step. The best among these will be chosen.

m

number of trees

nMin

minimum sample size for splitting a node

seed

Returns a list of RegressionTree objects which can be passed to predictRegression

def predictClassification(trees: Seq[ClassificationTree], samples: Mat[Double]): Mat[Double]
def predictRegression(trees: Seq[RegressionTree], samples: Mat[Double]): Vec[Double]