Skip to content

ADWIN

Adaptive Windowing method for concept drift detection1.

ADWIN (ADaptive WINdowing) is a popular drift detection method with mathematical guarantees. ADWIN efficiently keeps a variable-length window of recent items; such that it holds that there has no been change in the data distribution. This window is further divided into two sub-windows \((W_0, W_1)\) used to determine if a change has happened. ADWIN compares the average of \(W_0\) and \(W_1\) to confirm that they correspond to the same distribution. Concept drift is detected if the distribution equality no longer holds. Upon detecting a drift, \(W_0\) is replaced by \(W_1\) and a new \(W_1\) is initialized. ADWIN uses a significance value \(\delta=\in(0,1)\) to determine if the two sub-windows correspond to the same distribution.

Parameters

  • delta

    Default → 0.002

    Significance value.

  • clock

    Default → 32

    How often ADWIN should check for changes. 1 means every new data point, default is 32. Higher values speed up processing, but may also lead to increased delay in change detection.

  • max_buckets

    Default → 5

    The maximum number of buckets of each size that ADWIN should keep before merging buckets. The idea of data buckets comes from the compression algorithm introduced in the ADWIN2, the second iteration of the ADWIN algorithm presented in the original research paper. This is the ADWIN version available in River.

  • min_window_length

    Default → 5

    The minimum length allowed for a subwindow when checking for concept drift. Subwindows whose size is smaller than this value will be ignored during concept drift evaluation. Lower values may decrease delay in change detection but may also lead to more false positives.

  • grace_period

    Default → 10

    ADWIN does not perform any change detection until at least this many data points have arrived.

Attributes

  • drift_detected

    Whether or not a drift is detected following the last update.

  • estimation

    Estimate of mean value in the window.

  • n_detections

    The total number of detected changes.

  • total

    The sum of the stored elements.

  • variance

    The sample variance within the stored (adaptive) window.

  • width

    Window size

Examples

import random
from river import drift

rng = random.Random(12345)
adwin = drift.ADWIN()

data_stream = rng.choices([0, 1], k=1000) + rng.choices(range(4, 8), k=1000)

for i, val in enumerate(data_stream):
    _ = adwin.update(val)
    if adwin.drift_detected:
        print(f"Change detected at index {i}, input value: {val}")

Change detected at index 1023, input value: 4

Methods

update

Update the change detector with a single data point.

Apart from adding the element value to the window, by inserting it in the correct bucket, it will also update the relevant statistics, in this case the total sum of all values, the window width and the total variance.

Parameters

  • x — 'int | float'

Returns

DriftDetector: self

  1. Albert Bifet and Ricard Gavalda. "Learning from time-changing data with adaptive windowing." In Proceedings of the 2007 SIAM international conference on data mining, pp. 443-448. Society for Industrial and Applied Mathematics, 2007. ↩