Param Estimator

SeasonalityACF

Find candidate seasonality parameter using autocorrelation function CI.

Uses statsmodels.tsa.stattools.act for computing the autocorrelation function, and uses its testing functionality to determine candidate seasonality parameters. (“seasonality parameter” are integer lags, and abbreviated by sp, below)

Obtains confidence intervals at a significance level, and returns lags with significant positive auto-correlation, ordered by lower confidence limit.

Note: this should be applied to stationary series.: Quick stationarity transformation can be achieved by differencing. See also: Differencer

Schnellstart

python

from sktime.param_est.seasonality import SeasonalityACF

estimator = SeasonalityACF(candidate_sp=None, p_threshold=0.05, adjusted=False, nlags=None, fft=True, missing='none')

Parameter(6)

candidate_spNone, int or list of int, optional, default = None: candidate sp to test, and to restrict tests to; ints must be 2 or larger if None, will test all integer lags between 2 and nlags (inclusive)
p_thresholdfloat, optional, default=0.05: significance threshold to apply in testing for seasonality
adjustedbool, optional, default=False: If True, then denominators for autocovariance are n-k, otherwise n.
nlagsint, optional, default=None: Number of lags to compute autocorrelations for and select from. At default None, uses min(10 * np.log10(nobs), nobs - 1). Will be ignored if candidate_sp is provided.
fftbool, optional, default=True: If True, computes the ACF via FFT.
missingstr, [“none”, “raise”, “conservative”, “drop”], optional, default=”none”: Specifies how NaNs are to be treated. “none” performs no checks. “raise” raises an exception if NaN values are found. “drop” removes the missing observations and treats non-missing as contiguous. “conservative” computes the autocovariance using nan-ops so that nans are
removed when computing the mean and cross-products that are used to estimate the autocovariance. When using “conservative”, n is set to the number of non-missing observations.

Beispiele

>>> from sktime.datasets import load_airline
>>> from sktime.param_est.seasonality import SeasonalityACF
>>> 
>>> X = load_airline (). diff ()[1:]
>>> sp_est = SeasonalityACF ()
>>> sp_est. fit (X) SeasonalityACF(
... )
>>> sp_est. get_fitted_params ()["sp" ] 12
>>> sp_est. get_fitted_params ()["sp_significant" ] array([12, 11]) Series should be stationary before applying ACF. To pipeline SeasonalityACF with the Differencer, use the ParamFitterPipeline:
>>> from sktime.datasets import load_airline
>>> from sktime.param_est.seasonality import SeasonalityACF
>>> from sktime.transformations.difference import Differencer
>>> 
>>> X = load_airline ()
>>> sp_est = Differencer () * SeasonalityACF ()
>>> sp_est. fit (X) ParamFitterPipeline(
... )
>>> sp_est. get_fitted_params ()["sp" ] 12
>>> sp_est. get_fitted_params ()["sp_significant" ] array([12, 11])