Evaluation & Quality
ROC AUC
Area under the receiver operating characteristic curve
A score describing how well a system ranks positive cases above negative ones across different decision cutoffs.
Example
A test checks whether real cases generally receive higher scores than cases without the condition.
Why people use it
It summarizes how well scores separate two groups across possible cutoffs.
What you'll hear
“How well does it rank the real positives above the negatives?”
What this means for you
Look at mistakes at the cutoff you will actually use.
Can you control it?
No
No direct control. This describes a wider issue, concept or result rather than something you can simply switch on or off in a tool.
Common questions
- Does high ROC AUC guarantee useful precision on rare events?
- No. Precision also depends on prevalence and the chosen decision cutoff.
- Does it tell us which cutoff to use?
- No. Choosing when to act still depends on the consequences of different mistakes.
- Can it hide poor results at the cutoff we actually use?
- Yes. A good overall score can coexist with weak performance at a particular operating point.