d_overall: Determine overall desirability

In desirability2: Desirability Functions for Multiparameter Optimization

Determine overall desirability

Description

Once desirability columns have been created, determine the overall desirability using a mean (geometric by default).

Usage

d_overall(..., geometric = TRUE, tolerance = 0)

Arguments

...	One or more unquoted expressions separated by commas. To choose multiple columns using selectors, dplyr::across() can be used (see the example below).
geometric	A logical for whether the geometric or arithmetic mean should be used to summarize the columns.
tolerance	A numeric value where values strictly less than this value are capped at the value. For example, if users wish to use the geometric mean without completely excluding settings, a value greater than zero can be used.

Value

A numeric vector.

See Also

d_max()

Examples

library(dplyr)

# Choose model tuning parameters that minimize the number of predictors used
# while maximizing the area under the ROC curve.

classification_results %>%
  mutate(
    d_feat = d_min(num_features, 1, 200),
    d_roc  = d_max(roc_auc, 0.5, 0.9),
    d_all  = d_overall(across(starts_with("d_")))
  ) %>%
  arrange(desc(d_all))

# Bias the ranking toward minimizing features by using a larger scale.

classification_results %>%
  mutate(
    d_feat = d_min(num_features, 1, 200, scale = 3),
    d_roc  = d_max(roc_auc, 0.5, 0.9),
    d_all  = d_overall(across(starts_with("d_")))
  ) %>%
  arrange(desc(d_all))

desirability2 documentation built on May 31, 2023, 8:57 p.m.