Score — summary_score • renoir

Computes a score estimate given different sets of data, and returns the estimate and the relative standard error

Usage

summary_score(scorer, true, pred, weights, ...)

# S4 method for Scorer,list,list
summary_score(
  scorer,
  true,
  pred,
  weights = NULL,
  multi = c("average", "sum"),
  grouped = TRUE,
  min.obs = 3,
  logger,
  return.raw = FALSE,
  confidence = 0.95,
  distribution = "normal",
  ...
)

# S4 method for ScorerList,list,list
summary_score(
  scorer,
  true,
  pred,
  weights = NULL,
  multi = c("average", "sum"),
  grouped = TRUE,
  logger,
  ...
)

Arguments

scorer

a ScorerList object

true

a list of vectors (or matrices) of observed values. If list elements are matrices, a multi-response is assumed

pred

a list of vectors (or matrices) of predicted values

weights

a list of vectors of observation weights

...

further arguments to scorer function

multi

what to do when response has multiple output values

average: errors of multiple outputs are averaged to get a single value for each observation
sum: errors of multiple outputs are summed up to get a single value for each observation

grouped

logical, whether to compute separate statistics when lists of values are provided

min.obs

integer, minimum number of observations per list. If the number of observations per list is less than min.obs, an attempt to compute errors at the observation level (for the unlisted observations) and then summarise is made (equivalent to grouped = FALSE). It is working only for certain accuracy measures. Default is min.obs = 3. Set min.obs = 0 to suppress.

logger

a Logger object

return.raw

logical, whether to return scores used to compute estimate

Value

list of 2 elements containing a score estimate and its standard error

list containing the score estimate and its standard error for each Scorer in the ScorerList