Implementation:Online ml River Tree Nodes HTR

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

Overview

Hoeffding Tree Regressor (HTR) leaf node implementations providing different prediction strategies for regression tasks.

Description

This module provides three types of leaf nodes for Hoeffding Tree regressors. LeafMean predicts using the average target value and manages memory by removing bad splits from E-BST splitters. LeafModel uses a regression model (like linear regression) for predictions. LeafAdaptive dynamically selects between mean and model predictions based on which has lower FMSE (Faded Mean Squared Error).

Usage

Use LeafMean for simple predictions, LeafModel when a specific model is appropriate for the data, and LeafAdaptive when uncertain which strategy will perform better on your regression stream.

Code Reference

Source Location

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

Signature

class LeafMean(HTLeaf):
    def __init__(self, stats, depth, splitter, **kwargs):
        ...
    def prediction(self, x, *, tree=None):
        ...
    def manage_memory(self, criterion, last_check_ratio, last_check_vr, last_check_e):
        ...

class LeafModel(LeafMean):
    def __init__(self, stats, depth, splitter, leaf_model, **kwargs):
        ...
    def learn_one(self, x, y, *, w=1.0, tree=None):
        ...
    def prediction(self, x, *, tree=None):
        ...

class LeafAdaptive(LeafModel):
    def __init__(self, stats, depth, splitter, leaf_model, **kwargs):
        ...
    def learn_one(self, x, y, *, w=1.0, tree=None):
        ...
    def prediction(self, x, *, tree=None):
        ...

Import

from river.tree.nodes.htr_nodes import LeafMean
from river.tree.nodes.htr_nodes import LeafModel
from river.tree.nodes.htr_nodes import LeafAdaptive

I/O Contract

Input	Type	Description
stats	Var	Target variance statistics
depth	int	Node depth in tree
splitter	Splitter	Numeric attribute observer
leaf_model	Regressor	Model for predictions (LeafModel/LeafAdaptive only)

Output	Type	Description
prediction	float	Predicted target value
promise	int	Negative depth (for memory management)

Usage Examples

from river.tree.nodes.htr_nodes import LeafAdaptive
from river.tree.splitter import EBSTSplitter
from river import linear_model

# Create adaptive leaf with linear model
leaf = LeafAdaptive(
    stats=None,  # Var() will be created
    depth=2,
    splitter=EBSTSplitter(),
    leaf_model=linear_model.LinearRegression()
)

# Learn from samples
x = {'feature1': 0.5, 'feature2': 12.5}
leaf.learn_one(x, y=25.5, w=1.0, tree=None)

# Get prediction (chooses between mean and model)
pred = leaf.prediction(x, tree=None)

Related Pages

• Environment:Online_ml_River_Python_Runtime_Environment