📏 Partial AUC (pAUC)

Partial AUC (pAUC)#

AUC restricted to a region of interest of the ROC curve.

Important

✨ AI-generated content. This page was written with the assistance of an AI language model and is provided as a learning aid. Despite careful review, it may still contain mistakes, omissions, or out-of-date information. Whether you are new to the topic, a team lead, or a senior practitioner, treat it as a starting point rather than an authoritative reference: read it critically and independently verify anything you act on (code, commands, figures, and factual claims) against official documentation and primary sources before relying on it.

What it is#

Partial AUC is the area under only a slice of the ROC curve — typically a confined range of false positive rate (say \(\text{FPR} \le 0.1\)), sometimes of TPR, or both. It focuses the metric on the operating region that actually matters.

Why restrict#

Full AUC weights all FPR regions equally, but many are operationally irrelevant — a radiologist doesn’t care about performance at 80% FPR, a bank won’t run a fraud model that flags half of transactions. pAUC scores the model where it will be used (low FPR / high TPR for screening), and is especially apt for low-prevalence data needing high specificity.

The trade-off#

pAUC is more decision-relevant than full AUC and distinguishes curves that cross yet share the same total AUC, but it ignores the ROC outside the band and its raw value depends on the interval width (so it’s often standardized, e.g. McClish, back to \([0,1]\)). The bounds must be justified by the use case.

Theme: Classification & Averaging Metrics · All terminology

Hint

Mind map — connected ideas

ROC-AUC (Receiver Operating Characteristic – Area Under Curve, = AUROC) · Macro AUC · Micro AUC · Multiclass AUROC · Binary Classification · Precision–Recall AUC (PR-AUC)

Hint

More in Classification & Averaging Metrics

Accuracy · AUC (Area Under the Curve) · Average Precision (AP) · Binary Classification · Classification Probability · Discriminatory Power · F1-score · Gini Coefficient · Harmonic Mean · Log Loss (also called Logarithmic Loss or Cross-Entropy Loss) · Macro AUC · Macro AUROC (Macro-Averaged AUROC) · Macro Averaging · Macro F1