RankLags

class tsbootstrap.ranklags.RankLags(X: ndarray, model_type: Literal['ar', 'arima', 'sarima', 'var', 'arch'], max_lag: Integral = 10, y=None, save_models: bool = False)[source]

A class that uses several metrics to rank lags for time series models.

rank_lags_by_aic_bic()[source]: Rank lags based on Akaike information criterion (AIC) and Bayesian information criterion (BIC).

rank_lags_by_pacf()[source]: Rank lags based on Partial Autocorrelation Function (PACF) values.

estimate_conservative_lag()[source]: Estimate a conservative lag value by considering various metrics.

get_model(order)[source]: Retrieve a previously fitted model given an order.

Examples

>>> from tsbootstrap import RankLags
>>> import numpy as np
>>> X = np.random.normal(size=(100, 1))
>>> rank_obj = RankLags(X, model_type='ar')
>>> rank_obj.estimate_conservative_lag()
2
>>> rank_obj.rank_lags_by_aic_bic()
(array([2, 1]), array([2, 1]))
>>> rank_obj.rank_lags_by_pacf()
array([1, 2])

property X: ndarray

The input data.

Returns:: The input data.
Return type:: np.ndarray

estimate_conservative_lag() → int[source]

Estimate a conservative lag value by considering various metrics.

Returns:: A conservative lag value.
Return type:: int

get_model(order: int)[source]

Retrieve a previously fitted model given an order.

Parameters:: order (int) – Order of the model to retrieve.
Returns:: The fitted model.
Return type:: Union[AutoRegResultsWrapper, ARIMAResultsWrapper, SARIMAXResultsWrapper, VARResultsWrapper, ARCHModelResult]

property max_lag: Integral

Maximum lag to consider.

Returns:: Maximum lag to consider.
Return type:: int

property model_type: Literal['ar', 'arima', 'sarima', 'var', 'arch']

The type of model to fit.

Returns:: The type of model to fit.
Return type:: str

rank_lags_by_aic_bic()[source]

Rank lags based on Akaike information criterion (AIC) and Bayesian information criterion (BIC).

Returns:: aic_ranked_lags: Lags ranked by AIC. bic_ranked_lags: Lags ranked by BIC.
Return type:: Tuple[np.ndarray, np.ndarray]

rank_lags_by_pacf() → ndarray[source]

Rank lags based on Partial Autocorrelation Function (PACF) values.

Returns:: Lags ranked by PACF values.
Return type:: np.ndarray

property y: ndarray

Exogenous variables to include in the model.

Returns:: Exogenous variables to include in the model.
Return type:: np.ndarray