Knowledge Maxxing← Back to topic library
ML MaxxingIntermediate75 sec
ROC-AUC
ROC-AUC is the area under the ROC curve, a single number summarizing how well a model ranks positive examples above negative ones across all thresholds.
Prerequisite: ROC curve
What matters
Because it is threshold-independent, ROC-AUC is useful for comparing models before a specific operating threshold has been chosen for deployment.
Example
An ROC-AUC of 0.5 means the model ranks positives no better than random guessing, while 1.0 means it perfectly ranks every positive above every negative.
Common misconception
A high ROC-AUC does not guarantee good performance at the specific threshold actually used in production; it summarizes ranking quality across all thresholds, not one chosen operating point.
Recall question
What does an ROC-AUC of 0.5 tell you about a model's ranking ability?
Keywords and related concepts
ROC-AUCranking metricthreshold-independentroc curve