Zurück zu den Modellen
Detector

MovingWindow

API-ReferenzQuellcode

Moving window (MOSUM) changepoint detection algorithm.

Schnellstart

python
from sktime.detection.moving_window import MovingWindow

estimator = MovingWindow(change_score=None, penalty=None, bandwidth=20, selection_method='local_optimum', min_detection_fraction=0.2, local_optimum_fraction=0.4)

Parameter(6)

change_scoreBaseIntervalScorer, optional, default=CUSUM()
Change score (or cost, which is converted automatically).
penaltyfloat, np.ndarray, or None, default=None
Penalty value.
bandwidthint or list of int, default=20
Window half-width(s).
selection_methodstr, default=”local_optimum”

"local_optimum" or "detection_length".

min_detection_fractionfloat, default=0.2

Minimum detection interval fraction for "detection_length".

local_optimum_fractionfloat, default=0.4

Neighbourhood fraction for "local_optimum".

Beispiele

>>> from sktime.detection.moving_window import MovingWindow
>>> import numpy as np
>>> X = np. concatenate ([np. zeros (100), 10 * np. ones (100), np. zeros (100)])
>>> det = MovingWindow (bandwidth = 20, penalty = 20)
>>> det. fit (X). predict (X)

Referenzen

  1. [1 ] Eichinger, B. & Kirch, C. (2018). A MOSUM procedure for the estimation of multiple random change points. [2 ] Meier, A., Kirch, C. & Cho, H. (2021). mosum: A package for moving sums in change-point analysis. JSS, 97, 1-42.