reset_detector

Online reset change-point detector.

This module implements a detector that resets the online algorithm after each declared change point.

class pysatl_cpd.core.online.detectors.reset_detector.OnlineResetDetector(algorithm, *, threshold=nan, skip_period=0, max_runlength=None, collect_states=True, data_transformer=None)[source]

Bases: OnlineDetector, Generic

Online detector that resets its algorithm after every detected change point.

Iterates over a data provider, feeds each observation to the algorithm, compares the resulting statistic against a threshold, and optionally enforces a maximum run-length and a post-detection skip period.

Parameters:

  • algorithm (OnlineAlgorithm[TypeVar(DataT), TypeVar(ConfigurationT, bound= OnlineAlgorithmConfiguration), TypeVar(StateT, bound= OnlineAlgorithmState)]) – Algorithm instance that drives per-step detection.

  • threshold (float) – Detection threshold. A change point is declared when the algorithm’s detection statistic exceeds this value. Default is nan (no signal-based detections).

  • skip_period (int) – Number of steps to skip (suppress detections) after each declared change point. Must be non-negative.

  • max_runlength (int | None) – If not None, forces a change-point declaration once the run length exceeds this value. Must be positive if specified.

  • collect_states (bool) – Whether to collect algorithm state snapshots in step results. If False, algorithm_state will be None in all step results.

  • data_transformer (Optional[IDataTransformer[Any, Any]]) – Optional transformer applied to incoming data before detection.

Raises:

ValueError – If skip_period is negative, or if max_runlength is not None and not positive.

__init__(algorithm, *, threshold=nan, skip_period=0, max_runlength=None, collect_states=True, data_transformer=None)[source]
Parameters:
Return type:

None

property description: ChangePointDetectorDescription

Return a static description of this detector and its parameters.

Returns:

Description containing the detector name, threshold, skip period, max run-length, state-collection flag, and data transformer annotation.

Return type:

ChangePointDetectorDescription

clone()[source]

Create an independent copy of this detector.

The clone uses a recreated (fresh) algorithm instance and shares the same threshold, skip period, max run-length, and data transformer settings as the original.

Returns:

A new detector instance with identical configuration.

Return type:

Self