Param Estimator

SeasonalityPeriodogram

Score periodicities by their spectral power.

Computes seasonality periodogram based on iloc indices (not loc labels), and finds significant periods based on their spectral power, using Welch’s method of periodogram averaging [R7c18f055916d-1].

Computes significant periods based on a threshold of the maximum power, i.e., periods with power above thresh * maxpower are considered significant, and the one with highest power is considered the main seasonality period.

Significance is determined by thresholding as above, not by statistical testing.

Based on seasonal package by welch [R7c18f055916d-2].

Schnellstart

python

from sktime.param_est.seasonality import SeasonalityPeriodogram

estimator = SeasonalityPeriodogram(min_period=4, max_period=None, thresh=0.1)

Parameter(3)

min_periodint: Disregard periods shorter than this number of samples. Defaults to 4
max_periodint: Disregard periods longer than this number of samples. Defaults to None
threshfloat (0..1): Retain periods scoring above thresh*maxscore. Defaults to 0.10

Beispiele

>>> from sktime.datasets import load_airline
>>> from sktime.param_est.seasonality import SeasonalityPeriodogram
>>> X = load_airline (). diff ()[1:]
>>> sp_est = SeasonalityPeriodogram ()
>>> sp_est. fit (X) SeasonalityPeriodogram(
... )
>>> sp_est. get_fitted_params ()["sp" ] 6
>>> sp_est. get_fitted_params ()["sp_significant" ] array([6, 12, 14, 4, 10, 5])