Root Mean Squared Forecast Error

Usage

rmsfe(forecast, data)

Arguments

forecast: A forecast object as returned by forecast_model.
data: A tibble or data.frame containing the original data used for estimation.

Value

A tibble containing the root mean squared forecast error estimates.

Examples


specification <- dplyr::tibble(
 type = c(
   "n"
 ),
 dependent = c(
   "FinConsExpHH"
 ),
 independent = c(
   "FinConsExpGov + HICP_Gas"
 )
)

set.seed(123)
testdata <- dplyr::tibble(
 time = seq.Date(from = as.Date("2005-01-01"),
                 to = as.Date("2023-10-01"),
                 by = "quarter"),
 FinConsExpGov = rnorm(mean = 100, n = length(time)),
 HICP_Gas = rnorm(mean = 200, n = length(time)),
 FinConsExpHH  = 0.5 + 0.2*FinConsExpGov + 0.3 *
   HICP_Gas + rnorm(length(time), mean = 0, sd = 0.2))

testdata <- tidyr::pivot_longer(testdata,
                               cols = -time,
                               names_to = "na_item",
                               values_to = "values")

model <- run_model(specification = specification,
                  dictionary = dict,
                  inputdata_directory = testdata,
                  primary_source = "local",
                  present = FALSE,
                  quiet = TRUE,
                  saturation = "IIS")


insample_output <- forecast_insample(model, sample_share = 0.97)
#> [1] "Model Run 1 up to 2023-04-01"
#> [1] "Model Run 2 up to 2023-07-01"
#> [1] "Model Run 3 up to 2023-10-01"
#> [1] "Forecast 1 from 2023-04-01 to 2023-10-01"
#> [1] "Forecast 2 from 2023-07-01 to 2023-10-01"

insample_output$rmsfe
#> # A tibble: 4 × 5
#>   na_item      rmsfe start      end        method
#>   <chr>        <dbl> <date>     <date>     <chr> 
#> 1 Total RMSFE  0.194 2023-04-01 2023-10-01 AR    
#> 2 FinConsExpHH 0.194 2023-04-01 2023-10-01 AR    
#> 3 Total RMSFE  0.247 2023-07-01 2023-10-01 AR    
#> 4 FinConsExpHH 0.247 2023-07-01 2023-10-01 AR