com.eignex.kumulant/stat/regression/tree/RegressionTreeConfig

RegressionTreeConfig

@Serializable

data class RegressionTreeConfig(val delta: Double = 0.05, val deltaDecay: Double = 0.9, val tau: Double = 0.05, val minSamplesSplit: Double = 30.0, val minSamplesLeaf: Double = 5.0, val splitPeriod: Int = 10, val maxDepth: Int = 16, val maxNodes: Int = 1024, val metric: SplitMetric = VarianceReduction, val mtry: Int? = null)(source)

Tunables for RegressionTree growth, shared by DecisionTreeRegressionStat and RandomForestRegressionStat.

Constructors

RegressionTreeConfig

constructor(delta: Double = 0.05, deltaDecay: Double = 0.9, tau: Double = 0.05, minSamplesSplit: Double = 30.0, minSamplesLeaf: Double = 5.0, splitPeriod: Int = 10, maxDepth: Int = 16, maxNodes: Int = 1024, metric: SplitMetric = VarianceReduction, mtry: Int? = null)(source)

Properties

delta

val delta: Double(source)

Hoeffding-bound confidence threshold. Lower -> splits require more evidence.

deltaDecay

val deltaDecay: Double(source)

Multiplicative decay applied to delta per depth; slows growth near leaves.

maxDepth

val maxDepth: Int(source)

Hard ceiling on tree depth.

maxNodes

val maxNodes: Int(source)

Hard ceiling on internal + leaf nodes.

metric

val metric: SplitMetric(source)

Split criterion.

minSamplesLeaf

val minSamplesLeaf: Double(source)

Minimum weighted samples on each side of a candidate split.

minSamplesSplit

val minSamplesSplit: Double(source)

Minimum total weighted samples at a leaf before split evaluation.

mtry

val mtry: Int?(source)

Breiman-style random-subspace size: at every audit-leaf birth, draw a fresh random subset of this many candidates from the tree's full pool. null disables the trick.

splitPeriod

val splitPeriod: Int(source)

Audit every Nth observation rather than every update.

tau

val tau: Double(source)

If the Hoeffding bound itself shrinks below this, the leaf may split even when the runner-up is close (the classic VFDT "tie-break" parameter).