Implementation:Online ml River Stats Mean

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

Overview

Mean computes the running arithmetic mean of a data stream using Welford's algorithm.

Description

This statistic calculates the average value incrementally as data arrives, maintaining numerical stability through Welford's method. It supports weighted observations and includes a revert method for use with rolling windows. The implementation also provides batch update capabilities via update_many for efficient processing of numpy arrays. It can be combined with other statistics and supports parallel updates.

Usage

Use Mean when you need to track the average value in streaming data. This is one of the most fundamental statistics used in machine learning and data analysis. It's essential for normalization, baseline calculations, monitoring system metrics, and as a building block for more complex statistics like variance and covariance.

Code Reference

Source Location

• Repository: Online_ml_River
• File: river/stats/mean.py

Signature

class Mean(stats.base.Univariate):
    def __init__(self):
        self.n = 0
        self._mean = 0.0

class BayesianMean(stats.base.Univariate):
    def __init__(self, prior: float, prior_weight: float):
        self.prior = prior
        self.prior_weight = prior_weight
        self._mean = Mean()

Import

from river import stats

I/O Contract

Inputs

Name	Type	Required	Description
x	numbers.Number	Yes	Value to update the statistic with
w	float	No	Weight for the observation (default: 1.0)

Outputs

Name	Type	Description
get()	float	Current mean value

Usage Examples

from river import stats

# Basic running mean
X = [-5, -3, -1, 1, 3, 5]
mean = stats.Mean()

for x in X:
    mean.update(x)
    print(f"Value: {x}, Mean: {mean.get():.1f}")

# Output:
# Value: -5, Mean: -5.0
# Value: -3, Mean: -4.0
# Value: -1, Mean: -3.0
# Value: 1, Mean: -2.0
# Value: 3, Mean: -1.0
# Value: 5, Mean: 0.0

# Rolling mean using utils.Rolling
from river import utils

X = [1, 2, 3, 4, 5, 6]
rmean = utils.Rolling(stats.Mean(), window_size=2)

for x in X:
    rmean.update(x)
    print(f"Value: {x}, Rolling Mean: {rmean.get():.1f}")

# Output:
# Value: 1, Rolling Mean: 1.0
# Value: 2, Rolling Mean: 1.5
# Value: 3, Rolling Mean: 2.5
# Value: 4, Rolling Mean: 3.5
# Value: 5, Rolling Mean: 4.5
# Value: 6, Rolling Mean: 5.5

# Weighted mean
weighted_mean = stats.Mean()
weighted_mean.update(10, w=1.0)
weighted_mean.update(20, w=2.0)
weighted_mean.update(30, w=3.0)
print(f"Weighted mean: {weighted_mean.get():.2f}")
# Output: 23.33 (weighted towards higher values)

# Bayesian mean with prior
bayes_mean = stats.BayesianMean(prior=5.0, prior_weight=10)
for x in [1, 2, 3]:
    bayes_mean.update(x)
print(f"Bayesian mean: {bayes_mean.get():.2f}")
# Output incorporates prior belief

Related Pages

• Environment:Online_ml_River_Python_Runtime_Environment