TorchForecastingModel

This file contains several abstract classes:

• TorchForecastingModel is the super-class of all torch (deep learning) darts forecasting models.

• PastCovariatesTorchModel(TorchForecastingModel) for torch models consuming only past-observed covariates.

• FutureCovariatesTorchModel(TorchForecastingModel) for torch models consuming only future values of future covariates.

• DualCovariatesTorchModel(TorchForecastingModel) for torch models consuming past and future values of some single future covariates.

• MixedCovariatesTorchModel(TorchForecastingModel) for torch models consuming both past-observed as well as past and future values of some future covariates.

• SplitCovariatesTorchModel(TorchForecastingModel) for torch models consuming past-observed as well as future values of some future covariates.

• TorchParametricProbabilisticForecastingModel(TorchForecastingModel) is the super-class of all probabilistic torch forecasting models.

previous

Time-series Dense Encoder (TiDE)

next

Transformer Model