variable_effect function

Dataset Level Variable Effect as Partial Dependency Profile or Accumulated Local Effects

From DALEX version 1.0 this function calls the accumulated_dependence or partial_dependence from the ingredients package. Find information how to use this function here: https://ema.drwhy.ai/partialDependenceProfiles.html.

variable_effect(explainer, variables, ..., type = "partial_dependency")

variable_effect_partial_dependency(explainer, variables, ...)

variable_effect_accumulated_dependency(explainer, variables, ...)

Arguments

• explainer: a model to be explained, preprocessed by the 'explain' function
• variables: character - names of variables to be explained
• ...: other parameters
• type: character - type of the response to be calculated. Currently following options are implemented: 'partial_dependency' for Partial Dependency and 'accumulated_dependency' for Accumulated Local Effects

Returns

An object of the class 'aggregated_profiles_explainer'. It's a data frame with calculated average response.

Examples

titanic_glm_model <- glm(survived~., data = titanic_imputed, family = "binomial")
explainer_glm <- explain(titanic_glm_model, data = titanic_imputed)
expl_glm <- variable_effect(explainer_glm, "fare", "partial_dependency")
plot(expl_glm)

 
library("ranger")
titanic_ranger_model <- ranger(survived~., data = titanic_imputed, num.trees = 50,
                               probability = TRUE)
explainer_ranger  <- explain(titanic_ranger_model, data = titanic_imputed)
expl_ranger  <- variable_effect(explainer_ranger, variables = "fare",
                            type = "partial_dependency")
plot(expl_ranger)
plot(expl_ranger, expl_glm)

# Example for factor variable (with factorMerger)
expl_ranger_factor  <- variable_effect(explainer_ranger, variables =  "class")
plot(expl_ranger_factor)

References

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