tf_grouped_forecasts: Creates forecast for a dateframe with mutlitple timeseries...

In flostracke/efor: efor: EasyFORcasting, a package assisting forecasting multiple timeseries

Description

The Dataframe has to have the columny date, iterate and y. Following methods are supported:

Usage

tf_grouped_forecasts(data, n_pred, func, parallel = TRUE, ...)

Arguments

data	The dataframe containing the timeseries.
n_pred	The forecast horizon.
func	The used forecast method.
parallel	Specifies if the forecasts are created in parallel.
...	More arguments specific to the used forecasting method.

Details

• forecasts package: All methods from the forecast packages (https://CRAN.R-project.org/package=forecast) should work

• prophet: (https://CRAN.R-project.org/package=prophet)

• forecastHybrid: (https://CRAN.R-project.org/package=forecastHybrid)

• smooth package: (https://CRAN.R-project.org/package=smooth)

The forecasts are created in parallel with the help of the furrr package (https://github.com/DavisVaughan/furrr). Right now there is some bug with the prophet model in parallel.Thats why you should create prophet predictions with the parameter parallel = FALSE.

Value

The forecasted values for the dataset.

Examples

## Not run: tf_grouped_forecasts(
           train_data,        # used training dataset
           n_pred = 6,        # number of predictions
           func = auto.arima, # used forecasting method
           parallel = TRUE    # for runiing in parallel
         )
## End(Not run)

flostracke/efor documentation built on June 5, 2019, 5:36 p.m.