Forecaster

MultiplexForecaster

Categorical in XInsamplePred int insampleExogenous Multivariate

MultiplexForecaster for selecting among different models in Auto-ML pipelines.

MultiplexForecaster facilitates a framework for performing automated model selection process over different model classes. It should be used in conjunction with ForecastingGridSearchCV or similar tuners to build an Auto-ML pipeline for forecasters. MultiplexForecaster can be used with univariate and multivariate forecasters.

MultiplexForecaster is specified with a (named) list of forecasters and a selected_forecaster hyper-parameter, which is one of the forecaster names. The MultiplexForecaster then behaves precisely as the forecaster with name selected_forecaster, ignoring functionality in the other forecasters.

When used with ForecastingGridSearchCV, MultiplexForecaster provides an ability to tune across multiple estimators, i.e., to perform Auto-ML, by tuning the ,selected_forecaster, hyper-parameter. This combination will then select one of the passed forecasters via the tuning algorithm.

Quickstart

python

from sktime.forecasting.compose import MultiplexForecaster

estimator = MultiplexForecaster(forecasters: list, selected_forecaster=None)

Parameters(2)

forecasterslist of sktime forecasters, or: list of tuples (str, estimator) of sktime forecasters MultiplexForecaster can switch (“multiplex”) between these forecasters. These are “blueprint” forecasters, states do not change when fit is called.
selected_forecaster: str or None, optional, Default=None.: Name of the forecaster to be selected from the list of forecasters.
If str, must be one of the forecaster names. If no names are provided, must coincide with auto-generated name strings. To inspect auto-generated name strings, call get_params.
If None, behaves as if the first forecaster in the list is selected. Selects the forecaster as which MultiplexForecaster behaves.

Examples

>>> from sktime.forecasting.ets import AutoETS
>>> from sktime.forecasting.model_selection import ForecastingGridSearchCV
>>> from sktime.split import ExpandingWindowSplitter
>>> from sktime.forecasting.compose import MultiplexForecaster
>>> from sktime.forecasting.naive import NaiveForecaster
>>> from sktime.forecasting.theta import ThetaForecaster
>>> from sktime.forecasting.model_evaluation import evaluate
>>> from sktime.datasets import load_shampoo_sales
>>> y = load_shampoo_sales ()
>>> forecaster = MultiplexForecaster (forecasters = [
... ("ets", AutoETS ()),
... ("theta", ThetaForecaster ()),
... ("naive", NaiveForecaster ())])
>>> cv = ExpandingWindowSplitter (step_length = 12)
>>> gscv = ForecastingGridSearchCV (
... cv = cv,
... param_grid = { "selected_forecaster":["ets", "theta", "naive" ]},
... forecaster = forecaster)
>>> gscv. fit (y) ForecastingGridSearchCV(
... )