borg_pipeline: Validate an Entire Modeling Pipeline

In BORG: Bounded Outcome Risk Guard for Model Evaluation

Validate an Entire Modeling Pipeline

Description

Walks a tidymodels workflow() or caret::train() object and validates every step — preprocessing, feature selection, tuning, and model fitting — for information leakage.

Usage

borg_pipeline(pipeline, train_idx, test_idx, data = NULL, ...)

Arguments

pipeline	A modeling pipeline object. Supported types: A tidymodels workflow object (fitted or unfitted) A caret::train object A list with named components (recipe, model, tune_results, etc.)
train_idx	Integer vector of training row indices.
test_idx	Integer vector of test row indices.
data	Optional data frame. Required for parameter-level checks.
...	Additional arguments passed to inspectors.

Details

borg_pipeline() decomposes a pipeline into stages and inspects each:

1. Preprocessing: Recipe steps, preProcess, PCA, scaling

2. Feature selection: Variable importance, filtering

3. Hyperparameter tuning: Inner CV resamples

4. Model fitting: Training data scope, row counts

5. Post-processing: Threshold optimization, calibration

Each stage gets its own BorgRisk assessment. The overall result aggregates all risks across stages.

Value

An object of class "borg_pipeline" containing:

stages

Named list of per-stage BorgRisk results

overall

Aggregated BorgRisk for the full pipeline

n_stages

Number of stages inspected

leaking_stages

Character vector of stage names with hard violations

See Also

borg_validate, borg_inspect

Examples

if (requireNamespace("caret", quietly = TRUE)) {
  ctrl <- caret::trainControl(method = "cv", number = 5)
  model <- caret::train(mpg ~ ., data = mtcars[1:25, ], method = "lm",
                        trControl = ctrl, preProcess = c("center", "scale"))
  result <- borg_pipeline(model, train_idx = 1:25, test_idx = 26:32,
                          data = mtcars)
  print(result)
}

BORG documentation built on March 20, 2026, 5:09 p.m.