kumulant

TreeRegressionResult

@Serializable
@SerialName(value = "TreeRegressionResult")
data class TreeRegressionResult(val root: TreeNodeResult) : Result(source)

Immutable, wire-portable snapshot of a RegressionTree over a dense F64VectorView context. Carries the tree structure (SerializableSplit predicates + per-node weighted-variance aggregates) so callers can route a context vector to its leaf without reaching back into the live stat.

Serialization is only meaningful for the F64VectorView feature representation (the splits must be wire-portable); trees grown over other Split row types use the live RegressionTree directly and are not snapshotted to this type.

The root's WeightedVarianceResult is exposed as the canonical scalar snapshot; callers wanting context-specific predictions use findLeaf or predict.

Constructors

TreeRegressionResult

constructor(root: TreeNodeResult)(source)

Properties

root

Root of the snapshot tree.

rootMean

Root aggregate mean (over every observation absorbed).

totalWeights

Cumulative weight folded into the tree.

Functions

findLeaf

fun findLeaf(x: F64VectorView): WeightedVarianceResult(source)

Walk to the leaf the context resolves to.

predict

fun predict(x: F64VectorView): Double(source)

Mean of the leaf the context resolves to.