Detectors

Detectors operate on a single (univariate or multivariate) time series and segment it into homogeneous regions. All detectors inherit from BaseChangeDetector and expose the universal interface predict and predict_changepoints. Some detectors additionally expose predict_segment_anomalies, predict_scores or predict_all, depending on what the underlying algorithm computes.

Base

BaseChangeDetector()

Base class for all detectors providing sklearn compatibility.

Changepoint detectors

Detectors that implement predict and predict_changepoints.

CROPS([cost, min_penalty, max_penalty, ...])	CROPS algorithm for path solutions to the PELT algorithm.
MovingWindow([change_score, penalty, ...])	Moving window algorithm for multiple change-point detection.
PELT([cost, penalty, min_segment_length, ...])	Pruned exact linear time (PELT) changepoint detection.
SeededBinarySegmentation([change_score, ...])	Seeded binary segmentation algorithm for multiple changepoint detection.

Segment anomaly detectors

Detectors that additionally implement predict_segment_anomalies.

CAPA([segment_saving, point_saving, ...])	Collective and Point Anomaly (CAPA) detection algorithm.
CircularBinarySegmentation([...])	Circular binary segmentation algorithm for multiple segment anomaly detection.

Utilities

is_change_detector(estimator)

Return True if the given estimator is (probably) a change detector.