After passing a forecast object through the add_coverage function, it is re-transformed into a data.table.

[toshiakiasakura]

When I work on this package, I found the following bug.

As illustrated in the example, if I ran the following code, I obtained the "forecast_quantile" object.

example_quantile %>%
  set_forecast_unit(c("location", "target_end_date", "target_type", "horizon", "model")) %>%
  as_forecast() %>%
  class()
#> 'forecast_quantile''data.table''data.frame'

However, if I check class after passing through the add_coverage function, the object suddenly becomes a data.table.

example_quantile %>%
  set_forecast_unit(c("location", "target_end_date", "target_type", "horizon", "model")) %>%
  as_forecast() %>%
  add_coverage() %>% 
  class()
#> 'data.table''data.frame'

Expected outcome would be 'forecast_quantile''data.table''data.frame'.

[nikosbosse]

Adding to this, I think the same is true for transform_forecasts()