ingredients (version 2.2.0)

plot.feature_importance_explainer: Plots Feature Importance

Description

This function plots variable importance calculated as changes in the loss function after variable drops. It uses output from feature_importance function that corresponds to permutation based measure of variable importance. Variables are sorted in the same order in all panels. The order depends on the average drop out loss. In different panels variable contributions may not look like sorted if variable importance is different in different in different models.

Usage

# S3 method for feature_importance_explainer
plot(
  x,
  ...,
  max_vars = NULL,
  show_boxplots = TRUE,
  bar_width = 10,
  desc_sorting = TRUE,
  title = "Feature Importance",
  subtitle = NULL
)

Value

a ggplot2 object

Arguments

x

a feature importance explainer produced with the feature_importance() function

...

other explainers that shall be plotted together

max_vars

maximum number of variables that shall be presented for for each model. By default NULL what means all variables

show_boxplots

logical if TRUE (default) boxplot will be plotted to show permutation data.

bar_width

width of bars. By default 10

desc_sorting

logical. Should the bars be sorted descending? By default TRUE

title

the plot's title, by default 'Feature Importance'

subtitle

the plot's subtitle. By default - NULL, which means the subtitle will be 'created for the XXX model', where XXX is the label of explainer(s)

Details

Find more details in the Feature Importance Chapter.

References

Explanatory Model Analysis. Explore, Explain, and Examine Predictive Models. https://ema.drwhy.ai/

Examples

library("DALEX")
library("ingredients")

model_titanic_glm <- glm(survived ~ gender + age + fare,
                         data = titanic_imputed, family = "binomial")

explain_titanic_glm <- explain(model_titanic_glm,
                               data = titanic_imputed[,-8],
                               y = titanic_imputed[,8])

fi_rf <- feature_importance(explain_titanic_glm, B = 1)
plot(fi_rf)

# \donttest{
library("ranger")
model_titanic_rf <- ranger(survived ~., data = titanic_imputed, probability = TRUE)

explain_titanic_rf <- explain(model_titanic_rf,
                              data = titanic_imputed[,-8],
                              y = titanic_imputed[,8],
                              label = "ranger forest",
                              verbose = FALSE)

fi_rf <- feature_importance(explain_titanic_rf)
plot(fi_rf)

HR_rf_model <- ranger(status~., data = HR, probability = TRUE)

explainer_rf  <- explain(HR_rf_model, data = HR, y = HR$status,
                         verbose = FALSE, precalculate = FALSE)

fi_rf <- feature_importance(explainer_rf, type = "raw", max_vars = 3,
                            loss_function = DALEX::loss_cross_entropy)
head(fi_rf)
plot(fi_rf)

HR_glm_model <- glm(status == "fired"~., data = HR, family = "binomial")
explainer_glm <- explain(HR_glm_model, data = HR, y = as.numeric(HR$status == "fired"))

fi_glm <- feature_importance(explainer_glm, type = "raw",
                             loss_function = DALEX::loss_root_mean_square)
head(fi_glm)
plot(fi_glm)
# }