add_fc_models_to_main_forecasting_table: Add fc models to main forecasting table

In ing-bank/tsforecast: Time Series Forecasting Pipeline

Description

add_fc_models_to_main_forecasting_table is a function to extend the main forecasting table with an additional column, called fc_models, in which all the specified forecast methods are stored after they have been run. The forecast methods are run for each row of parameters and time series data in the main forecasting table. The resulting fc_models column consists of a named lists with forecast models and forecast data.

Usage

add_fc_models_to_main_forecasting_table(
  main_forecasting_table,
  periods_ahead = 12,
  fc_methods = supported_fc_methods_uni_var(),
  overwrite_fc = FALSE,
  add_fc_errors = TRUE,
  allow_negative_fc = FALSE,
  keep_fc_model_objects = FALSE,
  verbose = FALSE,
  parallel = TRUE,
  max_cores = Inf
)

Arguments

main_forecasting_table	A tibble containing several columns of data required for time series forecasting, which has been created using the create_main_forecasting_table function.
periods_ahead	A positive integer value indicating the number of periods to forecast ahead.
fc_methods	A character vector specifying the forecast methods to add. For more info `?supported_fc_methods`.
overwrite_fc	Boolean, which is set to TRUE if all existing forecasts should be overwritten, or set to FALSE if only unavailable forecasts should be added. Needs to be set to true if the @param main_forecasting_table already has forecasts of different @param periods_ahead.
add_fc_errors	Boolean, which is set to TRUE if add_fc_errors_to_main_forecasting_table should be run to add the fc_errors column as part of this function, or set to FALSE if it is not to be run (but to possibly be run separately later on).
allow_negative_fc	Boolean, which is to be set to TRUE if negative forecast values are allowed, or set to FALSE if negative forecast values should be overwritten by a value of zero.
keep_fc_model_objects	Boolean, which is set to TRUE in order to keep original fc_model objects in the main_forecasting_table after running the forecast. This is needed for scenario analysis in multivariate forecasting. However, it may lead to memory issues, as the main_forecasting_table increases in size.
verbose	Boolean, which is set to TRUE if status updates are valued, or set to FALSE if they are not.
parallel	Boolean, which is set to TRUE if each row of the main_forecast_table should be ran in parallel (using #logical cores - 1), or set to FALSE if they are not.
max_cores	A positive integer value indicating the maximum number of cores to use for running the forecasts in parallel.

Value

A tibble containing several columns of data required for time series forecasting, extended with as an additional column, called fc_models, with named lists with forecast models and forecast data.

Examples

data <- tstools::initialize_ts_forecast_data(
   data = dummy_gasprice,
   date_col = "year_month",
   col_of_interest = "gasprice",
   group_cols = c("state", "oil_company")
)
main_forecasting_table <- create_main_forecasting_table(
      data = data,
      seasonal_periods = c(12,3)
   ) %>%
   head(15)
main_forecasting_table <- add_fc_models_to_main_forecasting_table(
   main_forecasting_table,
   periods_ahead = 12,
   fc_methods = c("basic", "tree", "forest", "prophet")
)

ing-bank/tsforecast documentation built on Sept. 18, 2020, 9:40 a.m.