all_cutoffs: All cutoffs

In fairmodels: Flexible Tool for Bias Detection, Visualization, and Mitigation

All cutoffs

Description

Create all_cutoffs object and see how with the change of cutoffs parity loss of fairness metrics changes. Value of cutoff changes equally for all subgroups. User can pick which fairness metrics to create the object with via fairness_metrics vector.

Usage

all_cutoffs(
  x,
  grid_points = 101,
  fairness_metrics = c("ACC", "TPR", "PPV", "FPR", "STP")
)

Arguments

x	object of class fairness_object
grid_points	numeric, grid for cutoffs to test. Number of points between 0 and 1 spread evenly
fairness_metrics	character, name of parity_loss metric or vector of multiple metrics names. Full names can be found in fairness_check documentation.

Value

all_cutoffs object, data.frame containing information about label, metric and parity_loss at particular cutoff

Examples

data("german")

y_numeric <- as.numeric(german$Risk) - 1

lm_model <- glm(Risk ~ .,
  data = german,
  family = binomial(link = "logit")
)

explainer_lm <- DALEX::explain(lm_model, data = german[, -1], y = y_numeric)

fobject <- fairness_check(explainer_lm,
  protected = german$Sex,
  privileged = "male"
)

ac <- all_cutoffs(fobject)
plot(ac)

rf_model <- ranger::ranger(Risk ~ .,
  data = german,
  probability = TRUE,
  num.trees = 100,
  seed = 1
)


explainer_rf <- DALEX::explain(rf_model,
  data = german[, -1],
  y = y_numeric
)

fobject <- fairness_check(explainer_rf, fobject)

ac <- all_cutoffs(fobject)

plot(ac)

fairmodels documentation built on Aug. 24, 2022, 1:05 a.m.