Back to models
Forecaster

SkforecastRecursive

API referenceSource
Categorical in XPred intExogenous

Adapter for skforecast.recursive.ForecasterRecursive class [1].

Quickstart

python
from sktime.forecasting.compose import SkforecastRecursive

estimator = SkforecastRecursive(regressor: object, lags: int | list | ndarray | range | None=None, window_features: object | list | None=None, transformer_y: object | None=None, transformer_X: object | None=None, weight_func: Callable | None=None, differentiation: int | None=None, fit_kwargs: dict | None=None, binner_kwargs: dict | None=None, store_in_sample_residuals: bool=False)

Parameters(10)

regressorregressor or pipeline compatible with the scikit-learn API
An instance of a regressor or pipeline compatible with the scikit-learn API
lagsint, list, numpy ndarray, range, default None

Lags used as predictors. Index starts at 1, so lag 1 is equal to t-1.

  • int: include lags from 1 to lags (included).

  • list, 1d numpy ndarray or range: include only lags present in

lags, all elements must be int. - None: no lags are included as predictors.

window_featuresobject, list, default None

Instance or list of instances used to create window features. Window features are created from the original time series and are included as predictors. This argument is meant to work with RollingFeatures class [2].

transformer_yobject transformer (preprocessor), default None

An instance of a transformer (preprocessor) compatible with the scikit-learn preprocessing API with methods: fit, transform, fit_transform and inverse_transform. ColumnTransformer’s are not allowed since they do not have inverse_transform method. The transformation is applied to y before training the forecaster.

transformer_Xobject transformer (preprocessor), default None

An instance of a transformer (preprocessor) compatible with the scikit-learn preprocessing API. The transformation is applied to X before training the forecaster. inverse_transform is not available when using ColumnTransformer’s.

weight_funcCallable, default None

Function that defines the individual weights for each sample based on the index. For example, a function that assigns a lower weight to certain dates. Ignored if regressor does not have the argument sample_weight in its fit method. The resulting sample_weight cannot have negative values.

differentiationint, default None

Order of differencing applied to the time series before training the forecaster. If None, no differencing is applied. The order of differentiation is the number of times the differencing operation is applied to a time series. Differencing involves computing the differences between consecutive data points in the series. Differentiation is reversed in the output of predict() and predict_interval().

fit_kwargsdict, default None

Additional arguments to be passed to the fit method of the regressor.

binner_kwargsdict, default None

Additional arguments to pass to the QuantileBinner class [3] used to discretize the residuals into k bins according to the predicted values associated with each residual. Available arguments are:

  • n_bins

  • method

  • subsample

  • random_state

  • dtype

store_in_sample_residualsbool, default False

If True, stores the in-sample residuals when fitting the forecaster. This is required if you want to use predict_quantiles later. If False, predict_quantiles will raise an error unless you call set_in_sample_residuals() manually.

Argument method is passed internally to the function numpy.percentile.

Examples

>>> from sktime.forecasting.compose import SkforecastRecursive Without exogenous features
>>> from sklearn.linear_model import LinearRegression
>>> from sktime.datasets import load_airline
>>> y = load_airline ()
>>> forecaster = SkforecastRecursive (
... LinearRegression (), 2
... )
>>> forecaster. fit (y) SkforecastRecursive(lags=2, regressor=LinearRegression())
>>> y_pred = forecaster. predict (fh = [1, 2, 3 ])
>>> y_pred_int = forecaster. predict_interval (
... fh = [2 ], coverage = [0.9, 0.95 ]
... )
>>> y_pred_qtl = forecaster. predict_quantiles (
... fh = [1, 3 ], alpha = [0.8, 0.3, 0.2, 0.7 ]
... ) With exogenous features
>>> from sklearn.ensemble import RandomForestRegressor
>>> from sktime.datasets import load_longley
>>> y, X = load_longley ()
>>> y_train = y. head (n = 12)
>>> y_test = y. tail (n = 4)
>>> X_train = X. head (n = 12)
>>> X_test = X. tail (n = 4)
>>> forecaster = SkforecastRecursive (
... RandomForestRegressor (), [2, 4 ]
... )
>>> forecaster. fit (y_train, X = X_train) SkforecastRecursive(lags=[2, 4], regressor=RandomForestRegressor())
>>> y_pred = forecaster. predict (fh = [1, 2, 3 ], X = X_test)
>>> y_pred_int = forecaster. predict_interval (
... fh = [1, 3 ], X = X_test, coverage = [0.6, 0.4 ]
... )
>>> y_pred_qtl = forecaster. predict_quantiles (
... fh = [1, 3 ], X = X_test, alpha = [0.01, 0.5 ]
... )

References

  1. [1 ] https://skforecast.org/latest/api/forecasterrecursive#forecasterrecursive [2 ] https://skforecast.org/0.14.0/api/preprocessing#skforecast.preprocessing.preprocessing.RollingFeatures [3 ] https://skforecast.org/latest/api/preprocessing.html#skforecast.preprocessing.preprocessing.QuantileBinner