Skip to contents

Drop the heavy list columns of an allocated data frame

Source: R/slim.R
slim.Rd

An allocated object carries the full lambda-iteration history and a closure per target, which is far more than is needed to score many outcome vectors against the same allocation. slim() keeps the budgets, the allocations and the scores.

Usage

slim(
  adf,
  xdf_action = c("default", "unnest", "nest"),
  id_cols = NULL,
  rm_score_fun_if_scored = TRUE,
  rm_score_fun_if_not_scored = FALSE
)

Arguments

adf

an allocated object, as returned by allocate().

xdf_action

whether to unnest the xdf column. The default unnests an unscored allocation and leaves a scored one nested.

id_cols

additional columns to keep, such as a model name or origin time.

rm_score_fun_if_scored

drop the per-target scoring closure from a scored data frame, where it is usually no longer needed.

rm_score_fun_if_not_scored

drop the per-target scoring closure from an unscored data frame. Defaults to FALSE, since it is what alloscore.slim() scores with.

Value

A tibble of class slim, carrying the gpl_df, w and target_col_name attributes of adf.

Examples

fc <- add_pdqr_funs(
  tibble::tibble(target_names = c("a", "b"), dist = "norm", mean = c(5, 8), sd = 1),
  types = c("p", "q")
)
slim(allocate(fc, K = c(10, 20)))
#> # A tibble: 4 × 4
#>       K target_names     x score_fun
#> * <dbl> <chr>        <dbl> <list>   
#> 1    10 a             3.50 <fn>     
#> 2    10 b             6.50 <fn>     
#> 3    20 a             8.50 <fn>     
#> 4    20 b            11.5  <fn>