Implementation:Online ml River Stats AutoCorr

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

Overview

AutoCorr measures the serial correlation between the current value and a value seen n steps before.

Description

This statistic computes the Pearson correlation between the current value and the value observed at a specified lag. It maintains a sliding window to track past values and uses the PearsonCorr statistic to calculate the correlation coefficient. This is useful for detecting patterns and dependencies in time series data where current values may be related to previous observations.

Usage

Use AutoCorr when you need to detect autocorrelation in streaming time series data, such as identifying periodic patterns, seasonality, or temporal dependencies. Common applications include financial data analysis, sensor readings, and any sequential data where past values may influence current ones.

Code Reference

Source Location

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

Signature

class AutoCorr(stats.base.Univariate):
    def __init__(self, lag: int):
        self.window: collections.deque[numbers.Number] = collections.deque(maxlen=lag)
        self.lag = lag
        self.pearson = stats.PearsonCorr(ddof=1)

Import

from river import stats

I/O Contract

Inputs

Name	Type	Required	Description
x	numbers.Number	Yes	Value to update the statistic with
lag	int	Yes (init)	Number of steps to look back for correlation calculation

Outputs

Name	Type	Description
get()	float	Current autocorrelation coefficient (between -1 and 1)

Usage Examples

from river import stats

# Create autocorrelation with lag of 1
auto_corr = stats.AutoCorr(lag=1)

# Update with values
for x in [0.25, 0.5, 0.2, -0.05]:
    auto_corr.update(x)
    print(auto_corr.get())

# Output:
# 0
# 0
# -1.0
# 0.103552

# Different lag value
auto_corr = stats.AutoCorr(lag=2)
for x in [0.25, 0.5, 0.2, -0.05]:
    auto_corr.update(x)
    print(auto_corr.get())

# Output:
# 0
# 0
# 0
# -1.0

Related Pages

• Environment:Online_ml_River_Python_Runtime_Environment