Summarize SHAP values — summarize_contributions • powerup

Generate one consolidated matrix of aggregate SHAP contributions across POWERUP models. Models without SHAP contributions, including models skipped during training, are omitted.

Usage

summarize_contributions(
  models,
  use_new_data = FALSE,
  min_abs_contribution = 0.05
)

Arguments

models: A list of model objects generated by fit_powerup_models() and optionally updated with add_powerup_predictions().
use_new_data: If TRUE, summarize SHAP contributions from model$new_data; otherwise summarize training-data contributions.
min_abs_contribution: Minimum absolute aggregate contribution required for a feature to be retained in at least one model.

Value

A numeric matrix with models as rows and retained features as columns.

Examples

if (FALSE) { # \dontrun{
contributions <- summarize_contributions(my_models)
} # }