evaluate-model

Evaluate binary classification or regression models with confusion-matrix metrics, tie-aware ROC AUC, regression errors, and deterministic bootstrap confidence intervals; emphasizes held-out data, uncertainty, baselines, and subgroup checks.