Conditional SAGE

SAGE with conditional sampling (features are "marginalized" conditionally). Uses ConditionalARFSampler as default ConditionalSampler.

See also

MarginalSAGE

Super classes

xplainfi::FeatureImportanceMethod -> xplainfi::SAGE -> ConditionalSAGE

Public fields

sampler

(ConditionalSampler) Sampler for conditional marginalization.

Methods

Public methods

• ConditionalSAGE$new()

• ConditionalSAGE$clone()

Inherited methods

• xplainfi::FeatureImportanceMethod$importance()
• xplainfi::FeatureImportanceMethod$obs_loss()
• xplainfi::FeatureImportanceMethod$print()
• xplainfi::FeatureImportanceMethod$reset()
• xplainfi::FeatureImportanceMethod$scores()
• xplainfi::SAGE$compute()
• xplainfi::SAGE$plot_convergence()

---

Method new()

Creates a new instance of the ConditionalSAGE class.

Usage

ConditionalSAGE$new(
  task,
  learner,
  measure = NULL,
  resampling = NULL,
  features = NULL,
  n_permutations = 10L,
  sampler = NULL,
  batch_size = 5000L,
  n_samples = 100L,
  early_stopping = FALSE,
  se_threshold = 0.01,
  min_permutations = 10L,
  check_interval = 1L
)

Arguments

task, learner, measure, resampling, features, n_permutations, batch_size, n_samples, early_stopping, se_threshold, min_permutations, check_interval

Passed to SAGE.

sampler

(ConditionalSampler) Optional custom sampler. Defaults to ConditionalARFSampler.

---

Method clone()

The objects of this class are cloneable with this method.

Usage

ConditionalSAGE$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

Examples

library(mlr3)
task = tgen("friedman1")$generate(n = 100)

# \donttest{
# Using default ConditionalARFSampler (also handles all mixed data)
sage = ConditionalSAGE$new(
  task = task,
  learner = lrn("regr.ranger", num.trees = 50),
  measure = msr("regr.mse"),
  n_permutations = 3L,
  n_samples = 20
)
#> ℹ No <ConditionalSampler> provided, using <ConditionalARFSampler> with default settings.
#> ℹ No <Resampling> provided, using `resampling = rsmp("holdout", ratio = 2/3)`
#>   (test set size: 33)
sage$compute()
# }
# \donttest{
# For alternative conditional samplers:
custom_sampler = ConditionalGaussianSampler$new(
  task = task
)
sage_custom = ConditionalSAGE$new(
  task = task,
  learner = lrn("regr.ranger", num.trees = 50),
  measure = msr("regr.mse"),
  n_permutations = 5L,
  n_samples = 20,
  sampler = custom_sampler
)
#> ℹ No <Resampling> provided, using `resampling = rsmp("holdout", ratio = 2/3)`
#>   (test set size: 33)
sage_custom$compute()
# }