Implementation:Online ml River Compose Renamer

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

Overview

Feature naming transformers that rename, prefix, or suffix feature names in streaming data.

Description

This module provides three simple transformers for modifying feature names: Renamer, Prefixer, and Suffixer. These are pure transformers that operate only on feature names without modifying feature values.

Renamer applies a mapping dictionary to rename specific features. Keys in the mapping that don't exist in the input are silently ignored, making it safe to use with varying feature sets. The transformation modifies the dictionary keys in-place using pop and reassignment.

Prefixer and Suffixer add a fixed string to the beginning or end of all feature names respectively. These are useful for namespacing features from different sources or stages in a pipeline, helping to avoid name collisions when combining features from multiple transformers.

Usage

Use Renamer when you need to change specific feature names to match expected inputs of downstream components. Use Prefixer/Suffixer when combining features from multiple sources and you want to keep them clearly distinguished, or when you need to maintain feature provenance through a complex pipeline.

Code Reference

Source Location

Repository: Online_ml_River
File: river/compose/renamer.py

Signature

class Renamer(base.Transformer):
    def __init__(self, mapping: dict[str, str]):
        ...

class Prefixer(base.Transformer):
    def __init__(self, prefix: str):
        ...

class Suffixer(base.Transformer):
    def __init__(self, suffix: str):
        ...

Import

from river import compose

I/O Contract

Input (Renamer)
Parameter	Type	Description
mapping	dict[str, str]	Dictionary mapping old names to new names
x	dict	Feature dictionary to transform

Input (Prefixer)
Parameter	Type	Description
prefix	str	String to prepend to all feature names
x	dict	Feature dictionary to transform

Input (Suffixer)
Parameter	Type	Description
suffix	str	String to append to all feature names
x	dict	Feature dictionary to transform

Output
Method	Return Type	Description
transform_one(x)	dict	Dictionary with renamed features

Key Methods
Method	Parameters	Description
transform_one(x)	x: dict	Transforms feature names according to strategy

Usage Examples

from river import compose

# Renamer: Substitute specific feature names
mapping = {'a': 'alpha', 'c': 'gamma'}
x = {'a': 42, 'b': 12, 'c': 7}

renamer = compose.Renamer(mapping)
print(renamer.transform_one(x))
# {'alpha': 42, 'b': 12, 'gamma': 7}

# Keys not in mapping are unchanged
x2 = {'a': 42, 'b': 12}
print(renamer.transform_one(x2))
# {'alpha': 42, 'b': 12}

# Prefixer: Add prefix to all features
x = {'a': 42, 'b': 12}
prefixer = compose.Prefixer('feature_')
print(prefixer.transform_one(x))
# {'feature_a': 42, 'feature_b': 12}

# Suffixer: Add suffix to all features
suffixer = compose.Suffixer('_v1')
print(suffixer.transform_one(x))
# {'a_v1': 42, 'b_v1': 12}

# Use in pipeline to namespace features
from river import preprocessing
from river import feature_extraction
from river import stats

# Example: Combine features from different sources with prefixes
raw_scaler = compose.Prefixer('raw_') | preprocessing.StandardScaler()
agg_features = feature_extraction.Agg(
    on='value',
    by='category',
    how=stats.Mean()
)
agg_with_prefix = agg_features | compose.Prefixer('agg_')

# Combine both with union
combined = raw_scaler + agg_with_prefix

# Example: Rename features to match model expectations
feature_names = {
    'user_age': 'age',
    'user_gender': 'gender',
    'item_price': 'price'
}

pipeline = (
    compose.Renamer(feature_names) |
    preprocessing.StandardScaler()
)

# Example: Use suffix to track transformation stages
from river import linear_model

model = (
    preprocessing.StandardScaler() |
    compose.Suffixer('_scaled') |
    feature_extraction.PolynomialExtender(degree=2) |
    compose.Suffixer('_poly') |
    linear_model.LinearRegression()
)

# This creates feature names like:
# 'a_scaled', 'b_scaled', 'a_scaled*b_scaled_poly', etc.

# Example: Multiple renamers in sequence
rename_stage1 = compose.Renamer({'a': 'x', 'b': 'y'})
rename_stage2 = compose.Renamer({'x': 'feature1', 'y': 'feature2'})

pipeline = rename_stage1 | rename_stage2

x = {'a': 1, 'b': 2, 'c': 3}
print(pipeline.transform_one(x))
# {'feature1': 1, 'feature2': 2, 'c': 3}

# Example: Combine prefix and suffix
transformer = (
    compose.Prefixer('v2_') |
    compose.Suffixer('_normalized')
)

x = {'temperature': 20.5, 'humidity': 65}
print(transformer.transform_one(x))
# {'v2_temperature_normalized': 20.5, 'v2_humidity_normalized': 65}

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment