ROCAUC and Average Precision Metrics

[andrewdalpino]

Two Probabilistic metrics that measure the area under a curve. ROCAUC measure the areas under the ROC curve and Average Precision measure the area under the precision-recall curve.

[Copilot]

🟡 Changes recommended

All-positive Average Precision inputs and distinct numeric-string class names can produce incorrect scores.

2 open findings

• Handle all-positive labels correctly in average precision · New
• Use strict class-name comparison in ROC AUC and average precision · New

What changed in this PR

Adds ROC AUC and Average Precision to Rubix ML’s probabilistic cross-validation metrics.

Changes:

• Implements positive-class ranking metrics with tie handling.
• Adds tests for ranking quality, empty inputs, and integer classes.
• Documents both metrics and adds navigation links.

File	Description
tests/​CrossValidation/​Metrics/​ROCAUCTest.php	Tests ROC AUC scoring.
tests/​CrossValidation/​Metrics/​AveragePrecisionTest.php	Tests Average Precision scoring.
src/​CrossValidation/​Metrics/​ROCAUC.php	Implements rank-based ROC AUC.
src/​CrossValidation/​Metrics/​AveragePrecision.php	Implements recall-weighted precision.
mkdocs.yml	Adds metric navigation entries.
docs/​cross-validation/​metrics/​rocauc.md	Documents ROC AUC usage.
docs/​cross-validation/​metrics/​average-precision.md	Documents Average Precision usage.
docs/​cross-validation.md	Lists both probabilistic metrics.

🧠 Review effort: Balanced

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[samuel-akopyan]

Nice work, the ranking logic and tie handling look correct and I verified the expected values in the test providers by hand. A few small things below.

[apphp]

See my comments