Skip to contents

Run Complete XAI Pipeline

Source: R/AutoXAI.R
xai_pipeline.Rd

Runs the full explainability pipeline: importance, PDPs, SHAP, correlations.

Usage

xai_pipeline(
  learner,
  task,
  top_k = 10,
  n_shap_obs = 5,
  cor_threshold = 0.7,
  features = NULL,
  verbose = TRUE
)

Arguments

learner

A trained mlr3 learner

task

The mlr3 task

top_k

Number of top features to analyze in detail (default: 10)

n_shap_obs

Number of observations for SHAP (default: 5)

cor_threshold

Correlation threshold for warnings (default: 0.7)

features

Optional: pre-selected feature subset

verbose

Print progress (default: TRUE)

Value

A list containing all XAI results

Examples

if (FALSE) { # \dontrun{
if (requireNamespace("DALEX", quietly = TRUE) &&
    requireNamespace("ggplot2", quietly = TRUE)) {
  results <- xai_pipeline(learner, task, top_k = 10)
  plot(results$importance)
  print(results$correlations$high_cor_pairs)
}
} # }