Converts results from modeltime_accuracy() into either interactive (reactable) or static (gt) tables.

  .round_digits = 2,
  .sortable = TRUE,
  .show_sortable = TRUE,
  .searchable = TRUE,
  .filterable = FALSE,
  .expand_groups = TRUE,
  .title = "Accuracy Table",
  .interactive = TRUE,



A tibble that is the output of modeltime_accuracy()


Rounds accuracy metrics to a specified number of digits. If NULL, rounding is not performed.


Allows sorting by columns. Only applied to reactable tables. Passed to reactable(sortable).


Shows sorting. Only applied to reactable tables. Passed to reactable(showSortable).


Adds search input. Only applied to reactable tables. Passed to reactable(searchable).


Adds filters to table columns. Only applied to reactable tables. Passed to reactable(filterable).


Expands groups dropdowns. Only applied to reactable tables. Passed to reactable(defaultExpanded).


A title for static (gt) tables.


Return interactive or static tables. If TRUE, returns reactable table. If FALSE, returns static gt table.


Additional arguments passed to reactable::reactable() or gt::gt() (depending on .interactive selection).


A static gt table or an interactive reactable table containing the accuracy information.



The function respects dplyr::group_by() groups and thus scales with multiple groups.

Reactable Output

A reactable() table is an interactive format that enables live searching and sorting. When .interactive = TRUE, a call is made to reactable::reactable().

table_modeltime_accuracy() includes several common options like toggles for sorting and searching. Additional arguments can be passed to reactable::reactable() via ....

GT Output

A gt table is an HTML-based table that is "static" (e.g. non-searchable, non-sortable). It's commonly used in PDF and Word documents that does not support interactive content.

When .interactive = FALSE, a call is made to gt::gt(). Arguments can be passed via ....

Table customization is implemented using a piping workflow (%>%). For more information, refer to the GT Documentation.


library(tidyverse) library(lubridate) library(timetk) library(parsnip) library(rsample) library(modeltime) # Data m750 <- m4_monthly %>% filter(id == "M750") # Split Data 80/20 splits <- initial_time_split(m750, prop = 0.9) # --- MODELS --- # Model 1: prophet ---- model_fit_prophet <- prophet_reg() %>% set_engine(engine = "prophet") %>% fit(value ~ date, data = training(splits))
#> Disabling weekly seasonality. Run prophet with weekly.seasonality=TRUE to override this.
#> Disabling daily seasonality. Run prophet with daily.seasonality=TRUE to override this.
# ---- MODELTIME TABLE ---- models_tbl <- modeltime_table( model_fit_prophet ) # ---- ACCURACY ---- models_tbl %>% modeltime_calibrate(new_data = testing(splits)) %>% modeltime_accuracy() %>% table_modeltime_accuracy()