scale_by_forecaster: Scale error measures based on those of a particular...
In dshemetov/evalcast-mirror: Tools For Evaluating COVID Forecasters

Scale error measures based on those of a particular forecaster.

scale_by_forecaster(
  score_card,
  score_cols,
  base_forecaster_name,
  id_cols = c("forecaster", "ahead", "geo_value", "forecast_date", "data_source",
    "signal", "target_end_date", "incidence_period"),
  drop_base_entries = TRUE
)

`score_card`	score_card like that returned by `evaluate_predictions()`
`score_cols`	vector of column names in `score_card` to normalize
`base_forecaster_name`	name of forecaster in `score_card$forecaster` column by whose error values the remaining forecasters' errors will be scaled
`id_cols`	vector of column names in `score_card` that identify distinct forecasts (i.e. the independent variables of `score_card`).
`drop_base_entries`	whether to drop the entries in `score_card` from `base_forecaster_name` (when FALSE their corresponding `score_cols` values will all be 1)

A tibble whose columns are c(id_cols, score_cols) whose id_cols values are copied directly from score_card and whose score_cols values are normalized with respect to base_forecaster_name.

dshemetov/evalcast-mirror documentation built on Feb. 4, 2022, 8:52 a.m.