Summarize models — summarize

Summarize model performance and metadata.

Usage

summarize_models(models)

A data frame with one row per model and columns:

brd: Model or perturbation identifier.
skipped: Whether the response model was not retained.
r: Mean held-out Pearson correlation across cross-validation folds.
R2: Mean held-out R-squared across cross-validation folds.
rmse: Mean held-out root mean squared error across cross-validation folds.
d_sensitivity: Mean held-out sensitivity after applying the response cutoff.
d_specificity: Mean held-out specificity after applying the response cutoff.
d_FPR: Mean held-out false-positive rate after applying the response cutoff.
d_PPV: Mean held-out positive predictive value after applying the response cutoff.
d_NPV: Mean held-out negative predictive value after applying the response cutoff.
d_accuracy: Mean held-out classification accuracy after applying the response cutoff.
n_terms: Number of fitted features with nonzero aggregate SHAP contribution across the training samples.
mean_pred_sd: Mean training-sample prediction SD estimated by the uncertainty model, when available.

if (FALSE) { # \dontrun{
model_summary <- summarize_models(my_models)
} # }