Implementation:Online ml River Stats Sum

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

Overview

Sum computes the running sum of values in a data stream.

Description

This statistic maintains a cumulative sum of all values observed in the stream. It is one of the simplest statistics and updates in constant time O(1). The implementation includes a revert method, allowing it to be used with rolling windows to compute windowed sums. It serves as a building block for many other statistics and aggregations.

Usage

Use Sum when you need to track the cumulative total of streaming values. Common applications include counting events, aggregating quantities (sales, revenue, costs), monitoring total resource usage, accumulating errors or scores, and as a component in calculating other statistics like mean (sum/count) or moving averages.

Code Reference

Source Location

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

Signature

class Sum(stats.base.Univariate):
    def __init__(self):
        self.sum = 0.0

Import

from river import stats

I/O Contract

Inputs

Name	Type	Required	Description
x	numbers.Number	Yes	Value to add to the running sum

Outputs

Name	Type	Description
get()	float	Current cumulative sum

Usage Examples

from river import stats

# Basic running sum
X = [-5, -3, -1, 1, 3, 5]
total = stats.Sum()

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

# Output:
# Value: -5, Sum: -5.0
# Value: -3, Sum: -8.0
# Value: -1, Sum: -9.0
# Value: 1, Sum: -8.0
# Value: 3, Sum: -5.0
# Value: 5, Sum: 0.0

# Rolling sum
from river import utils

X = [1, -4, 3, -2, 2, 1]
rolling_sum = utils.Rolling(stats.Sum(), window_size=2)

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

# Output:
# Value: 1, Rolling Sum: 1.0
# Value: -4, Rolling Sum: -3.0
# Value: 3, Rolling Sum: -1.0
# Value: -2, Rolling Sum: 1.0
# Value: 2, Rolling Sum: 0.0
# Value: 1, Rolling Sum: 3.0

# Accumulating sales revenue
revenue_sum = stats.Sum()
sales = [100, 250, 175, 300, 225]

for sale in sales:
    revenue_sum.update(sale)
    print(f"Sale: ${sale}, Total Revenue: ${revenue_sum.get():.2f}")

# Output:
# Sale: $100, Total Revenue: $100.00
# Sale: $250, Total Revenue: $350.00
# Sale: $175, Total Revenue: $525.00
# Sale: $300, Total Revenue: $825.00
# Sale: $225, Total Revenue: $1050.00

# Using sum with count to compute mean
total = stats.Sum()
count = stats.Count()

data = [10, 20, 30, 40, 50]
for x in data:
    total.update(x)
    count.update()
    mean = total.get() / count.get()
    print(f"Value: {x}, Mean so far: {mean:.1f}")

# Monitoring cumulative error
error_sum = stats.Sum()
predictions = [2.1, 3.8, 5.2, 4.9]
actuals = [2.0, 4.0, 5.0, 5.0]

for pred, actual in zip(predictions, actuals):
    error = pred - actual
    error_sum.update(error)
    print(f"Pred: {pred}, Actual: {actual}, Cumulative Error: {error_sum.get():.2f}")

Related Pages

• Environment:Online_ml_River_Python_Runtime_Environment