calculate_forecast_errors: Calculate forecast errors for fc_models.

In ing-bank/tsforecast: Time Series Forecasting Pipeline

Description

calculate_forecast_errors is a function to combine the fc_data from each forecast model within the named list fc_models and combine it with actuals data from the validation time series object, in order to calculate the forecast errors.

Usage

calculate_forecast_errors(
  fc_models,
  ts_object_train,
  ts_object_valid,
  allow_negative_fc = F
)

Arguments

fc_models	A named list of forecast models, with for each forecast model a list with the model itself and a table with forecast values.
ts_object_train	A time series object, which contains only the training data.
ts_object_valid	A time series object, which contains only the validation data.
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.

Value

A tibble with data on the fc errors for all the forecast models in fc_models.

Examples

ts_object_both <- tstools::initialize_ts_forecast_data(
      data = dummy_gasprice,
      date_col = "year_month",
      col_of_interest = "gasprice",
      group_cols = c("state", "oil_company")
   ) %>%
   dplyr::filter(grouping == "state = New York   &   oil_company = CompanyA") %>%
   tstools::transform_data_to_ts_object() %>%
   split_ts_object(ts_split_date = 200301)
calculate_forecast_errors(
   fc_models = add_all_univariate_forecast_models(
      ts_object_train = ts_object_both$train,
      periods_ahead = 12,
      fc_methods = c("basic", "linear", "prophet")
   ),
   ts_object_train = ts_object_both$train,
   ts_object_valid = ts_object_both$valid
)

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