Implementation:Online ml River Tree Nodes HATR

Knowledge Sources	Online_ml_River
Domains	Online_Learning, Decision_Trees
Last Updated	2026-02-08 16:00 GMT

Overview

Hoeffding Adaptive Tree Regressor (HATR) node implementations with drift detection and alternate subtree management for regression tasks.

Description

This module implements adaptive nodes for Hoeffding Adaptive Tree regressors. The AdaLeafRegressor monitors prediction errors using variance statistics and drift detectors. When drift is detected in branch nodes (AdaBranchRegressor), alternate subtrees are grown and evaluated using statistical tests (z-tests) to determine if they should replace the main subtree. Supports mean prediction, model-based prediction, and adaptive prediction strategies.

Usage

Use HATR nodes when building regression trees that must adapt to concept drift in non-stationary data streams. The statistical testing approach ensures reliable subtree replacement decisions.

Code Reference

Source Location

• Repository: Online_ml_River
• File: river/tree/nodes/hatr_nodes.py

Signature

class AdaLeafRegressor(HTLeaf):
    def __init__(self, stats, depth, splitter, drift_detector, rng, **kwargs):
        ...
    def learn_one(self, x, y, *, w=1.0, tree=None, parent=None, parent_branch=None):
        ...

class AdaBranchRegressor(DTBranch):
    def __init__(self, stats, *children, drift_detector, **attributes):
        ...
    def learn_one(self, x, y, *, w=1.0, tree=None, parent=None, parent_branch=None):
        ...
    def kill_tree_children(self, tree):
        ...

class AdaLeafRegMean(AdaLeafRegressor, LeafMean):
    ...

class AdaLeafRegModel(AdaLeafRegressor, LeafModel):
    ...

class AdaLeafRegAdaptive(AdaLeafRegressor, LeafAdaptive):
    ...

Import

from river.tree.nodes.hatr_nodes import AdaLeafRegressor
from river.tree.nodes.hatr_nodes import AdaBranchRegressor
from river.tree.nodes.hatr_nodes import AdaLeafRegMean

I/O Contract

Input	Type	Description
x	dict	Feature dictionary
y	float	Target value
w	float	Sample weight (default 1.0)
drift_detector	DriftDetector	Drift detection algorithm instance

Output	Type	Description
prediction	float	Predicted target value
error_tracker	Var	Variance statistics of prediction errors

Usage Examples

from river.tree.nodes.hatr_nodes import AdaLeafRegMean
from river.drift import ADWIN
from river.tree.splitter import EBSTSplitter
import random

# Create adaptive regression leaf
rng = random.Random(42)
leaf = AdaLeafRegMean(
    stats=None,  # Var() will be created
    depth=1,
    splitter=EBSTSplitter(),
    drift_detector=ADWIN(),
    rng=rng
)

# Update with samples
x = {'feature1': 0.5, 'feature2': 1.2}
leaf.learn_one(x, y=25.5, w=1.0, tree=None)

# Get prediction (mean value)
pred = leaf.prediction(x, tree=None)

Related Pages

• Environment:Online_ml_River_Python_Runtime_Environment