Core Module Documentation#

neuralprophet.components.router.get_future_regressors(config, id_list, quantiles, n_forecasts, device, config_trend_none_bool)#

Router for all future regressor classes.

neuralprophet.components.router.get_seasonality(config, id_list, quantiles, num_seasonalities_modelled, num_seasonalities_modelled_dict, n_forecasts, device)#

Router for all seasonality classes.

neuralprophet.components.router.get_trend(config, n_forecasts, quantiles, id_list, num_trends_modelled, device)#

Router for all trend classes.

Based on the conditions provided, the correct trend class is returned and initialized using the provided args.


config : configure.Trend n_forecasts : int

number of steps to forecast. Aka number of model outputs


the set of quantiles estimated


List of different time series IDs, used for global-local modelling (if enabled)

This parameter is set to ['__df__'] if only one time series is input.


Number of different trends modelled.

If only 1 time series is modelled, it will be always 1.

For multiple time series. If trend is modelled globally the value is set to 1, otherwise it is set to the number of time series modelled.


Device that tensors are stored on.

This is set to torch.device("cpu") if no GPU is available.