library("xai2shiny")

Titanic dataset

DALEX packages provides an imputed version of a common classification-oriented dataset - titanic (data was copied from the stablelearner package). Let’s see a sample of observations:

library("DALEX")
head(titanic_imputed, 3)

#>   gender age class    embarked  fare sibsp parch survived
#> 1   male  42   3rd Southampton  7.11     0     0        0
#> 2   male  13   3rd Southampton 20.05     0     2        0
#> 3   male  16   3rd Southampton 20.05     1     1        0

Models to compare

Package allows to pass multiple models from different packages to the main function, so why not create some:

library("ranger")

model_rf <- ranger(survived ~ .,
                   data = titanic_imputed,
                   classification = TRUE,
                   probability = TRUE)
model_glm <- glm(survived ~ .,
                 data = titanic_imputed,
                 family = "binomial")

Explainers based on models

In fact, xai2shiny function accepts only explainers, i.e. DALEX special objects basing on provided models. Let’s create all necessary explainers:

explainer_rf <- explain(model_rf,
                        data = titanic_imputed[,-8],
                        y = titanic_imputed$survived)
explainer_glm <- explain(model_glm,
                         data = titanic_imputed[,-8],
                         y = titanic_imputed$survived)

Shiny application

After that, the only thing left to do is to generate an app and run it:

xai2shiny(explainer_rf, explainer_glm, directory = './', run = TRUE)

Cloud deployment

Further cloud deployment can be done in 2 simple steps (see README example for details):

xai2shiny::cloud_setup()
my_droplet_id <- 1 # Compare it to your DigitalOcean account and set a proper ID 
deploy_shiny(droplet = my_droplet_id, directory = './xai2shiny', packages = "ranger")

External package’s model

As xai2shiny covers as many external models sources as DALEX and DALEXtra, let’s consider widely known mlr3 package:

library("DALEXtra")
library("mlr3")
library("mlr3learners")

titanic <- titanic_imputed
titanic[, 'survived'] <- as.factor(titanic[, 'survived'])

task <- TaskClassif$new(id = 'titanic', 
                        backend = titanic, 
                        target = "survived", 
                        positive = '1')

learner <- mlr_learners$get('classif.log_reg')
learner$predict_type = "prob"

train_set = sample(task$nrow, 0.8 * task$nrow)
test_set = setdiff(seq_len(task$nrow), train_set)

learner$train(task, row_ids = train_set)
explainer_mlr <- explain_mlr3(learner,
                              data = titanic[,-8], 
                              y = as.numeric(as.character(titanic$survived)),
                              label = "mlr3 model")
xai2shiny(explainer_mlr, directory = "./", run = FALSE)

xai2shiny introduction: Titanic dataset with multiple models