kats.models.ensemble.ensemble module

Ensemble techniques for forecasting

This implements a set of ensemble techniques including weighted averaging, median ensemble and STL-based ensembling method. This is the parent class for all ensemble models.

class kats.models.ensemble.ensemble.BaseEnsemble(data: kats.consts.TimeSeriesData, params: kats.models.ensemble.ensemble.EnsembleParams)

Bases: object

Base ensemble class

Implement parent class for ensemble.

fit()

Fit method for ensemble model

This method fits each individual model for ensembling and create a dict of model and fitted obj, such as {‘m1’: fitted_m1_obj, ‘m2’: fitted_m2_obj}

plot()

Plot method for ensemble model (not implemented yet)

class kats.models.ensemble.ensemble.BaseModelParams(model_name: str, model_params: kats.models.model.Model, **kwargs)

Bases: object

Ensemble parameter class

This class contains three attributes:

model

model names (str)

model_params

model_param is defined in base models