module change_detection.base_change_detector

Base Change Detector.

This module encapsulates functionality for global and partial (i.e. feature-wise) concept drift detection. The abstract BaseChangeDetector class should be used as super class for all concept drift detection methods.

Copyright (C) 2022 Johannes Haug.

class BaseChangeDetector

Abstract base class for change detection models.

Attributes:

  • reset_after_drift (bool): A boolean indicating if the change detector will be reset after a drift was detected.
  • error_based (bool): A boolean indicating if the change detector partial_fit function requires error measurements (i.e., a boolean vector indicating correct predictions) as input from the pipeline. This is true for most change detectors. If the attribute is False, the partial_fit function will receive raw input observations and targets from the pipeline (e.g., required by ERICS).
  • drifts (list): A list of time steps corresponding to detected concept drifts.
  • partial_drifts (List[tuple]): A list of time steps and features corresponding to detected partial concept drifts. A partial drift is a concept drift that is restricted to one or multiple (but not all) input features. Some change detectors are able to detect partial concept drift. This attribute is a list of tuples of the form (time step, [features under change]).
  • warnings (list): A list of time steps corresponding to warnings. Some change detectors are able to issue warnings before an actual drift alert.

method BaseChangeDetector.__init__

__init__(reset_after_drift: bool, error_based: bool)

Inits the change detector.

Args:

  • reset_after_drift: A boolean indicating if the change detector will be reset after a drift was detected.
  • error_based: A boolean indicating if the change detector partial_fit function requires error measurements (i.e., a boolean vector indicating correct predictions) as input from the pipeline. This is true for most change detectors. If the attribute is False, the partial_fit function will receive raw input observations and targets from the pipeline (e.g., required by ERICS).

method BaseChangeDetector.detect_change

detect_change() → bool

Detects global concept drift.

Returns:

  • bool: True, if a concept drift was detected, False otherwise.

method BaseChangeDetector.detect_partial_change

detect_partial_change() → Tuple[bool, list]

Detects partial concept drift.

Returns:

  • bool: True, if at least one partial concept drift was detected, False otherwise.
  • list: Indices (i.e. relative positions in the feature vector) of input features with detected partial drift.

method BaseChangeDetector.detect_warning_zone

detect_warning_zone() → bool

Detects a warning zone.

Some change detectors issue warnings before the actual drift alert.

Returns:

  • bool: True, if the change detector has detected a warning zone, False otherwise.

method BaseChangeDetector.partial_fit

partial_fit(*args, **kwargs)

Updates the change detector.

method BaseChangeDetector.reset

reset()

Resets the change detector.

This file was automatically generated via lazydocs.