lares (version 4.8.4)

dalex_explainer: DALEX Explainer for H2O

Description

DALEX function to create an explainer object

Usage

dalex_explainer(df, model, y = "tag", ignore = NA)

Arguments

df

Dataframe. Must contain all columns and predictions

model

Model object (H2O)

y

Character. Variable's column name

ignore

Character vector. Which columns should be ignored?

See Also

Other Interpretability: dalex_local(), dalex_residuals(), dalex_variable()

Examples

# NOT RUN {
data(dft) # Titanic dataset

# TRAIN A SIMPLE MODEL
dfm <- h2o_automl(dft, y = "Survived",
                  ignore = c("Ticket", "PassengerId", "Cabin"),
                  max_models = 1)

# EXPLAINER
explainer <- dalex_explainer(df = dfm$datasets$test, model = dfm$model, y = "Survived")
explainer$data <- na.omit(explainer$data)

# CATEGORICAL EXAMPLE
class <- dalex_variable(explainer, vars = c("Pclass", "Sex"))
class$plot

# NUMERICAL EXAMPLE
num <- dalex_variable(explainer, vars = c("Fare", "Age"))
num$plot

# LOCAL EXAMPLE
local <- dalex_local(explainer, row = 1)
# OR YOU COULD MANUALLY INPUT THE OBSERVATION
local <- dalex_local(explainer, observation = explainer$data[1,])
local$plot
# }