Estimate a Provenance-Only Prediction Floor — os_provenance_floor

Fits simple train-fold-only logistic provenance models to estimate whether non-expression source variables alone can predict the outcome. This is a lightweight package-level governance helper, not a replacement for a full prespecified modeling script.

Usage

os_provenance_floor_suite(
  data,
  outcome,
  provenance_predictors,
  group_id = NULL,
  positive = NULL,
  n_folds = 5L,
  seed = 1L
)

Arguments

data: Data frame containing outcome and provenance predictors.
outcome: Column name for the binary outcome.
provenance_predictors: Character vector of non-expression predictor columns.
group_id: Optional column name or vector defining grouped folds.
positive: Optional positive-class label.
n_folds: Number of grouped folds.
seed: Random seed.

Value

An os_provenance_floor object.