Back to models
Param Estimator

ARLagOrderSelector

API referenceSource

Estimate optimal lag order for autoregressive models using information criteria.

Quickstart

python
from sktime.param_est.lag import ARLagOrderSelector

estimator = ARLagOrderSelector(maxlag, ic='bic', glob=False, trend='c', seasonal=False, hold_back=None, period=None, missing='none')

Parameters(8)

maxlagint
Maximum number of lags to consider
icstr, default=”bic”

Information criterion to use for model selection:

  • “aic”: Akaike Information Criterion

  • “bic”: Bayesian Information Criterion (default)

  • “hqic”: Hannan-Quinn Information Criterion

globbool, default=False
If True, searches globally across all lag combinations up to maxlag. If False, searches sequentially by adding one lag at a time.
trendstr, default=”c”

Trend to include in the model:

  • “n”: No trend

  • “c”: Constant only

  • “t”: Time trend only

  • “ct”: Constant and time trend

seasonalbool, default=False
Whether to include seasonal dummies in the model
hold_backint, optional (default=None)
Number of initial observations to exclude from the estimation sample
periodint, optional (default=None)
Period of the data (used only if seasonal=True)
missingstr, default=”none”

How to handle missing values:

  • “none”: No handling

  • “drop”: Drop missing observations

  • “raise”: Raise an error

Examples

>>> from sktime.datasets import load_airline
>>> from sktime.param_est.lag import ARLagOrderSelector
>>> y = load_airline ()
>>> selector = ARLagOrderSelector (maxlag = 12, ic = "bic")
>>> selector. fit (y) ARLagOrderSelector(
... )
>>> selector. selected_model_ (3,)
>>> selector. ic_value_ 1369.6963340649502