fairness_heatmap: Fairness heatmap

In ModelOriented/FairModels: Flexible Tool for Bias Detection, Visualization, and Mitigation

Fairness heatmap

Description

Create fairness_heatmap object to compare both models and metrics. If parameter scale is set to TRUE metrics will be scaled to median = 0 and sd = 1. If NA's appear heatmap will still plot, but with gray area where NA's were.

Usage

fairness_heatmap(x, scale = FALSE)

Arguments

x	object of class fairness_object
scale	logical, if codeTRUE metrics will be scaled to mean 0 and sd 1. Default FALSE

Value

fairness_heatmap object.

It is a list with following fields:

• heatmap_data - data.frame with information about score for model and parity loss metric

• matrix_model - matrix used in dendogram plots

• scale - logical parameter passed to fairness_heatmap

• label - character, vector of model labels

Examples

data("german")

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

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

rf_model <- ranger::ranger(Risk ~ .,
  data = german,
  probability = TRUE,
  num.trees = 200,
  num.threads = 1
)

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

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

# same explainers with different cutoffs for female
fobject <- fairness_check(explainer_lm, explainer_rf, fobject,
  protected = german$Sex,
  privileged = "male",
  cutoff = list(female = 0.4),
  label = c("lm_2", "rf_2")
)


fh <- fairness_heatmap(fobject)

plot(fh)

ModelOriented/FairModels documentation built on Jan. 31, 2025, 8:16 p.m.