ROC AUC Calculator

Area under the ROC curve via the trapezoidal rule over descending scores; equal scores share one threshold point so ties are handled exactly.

Description

Calculate binary receiver-operating-characteristic area under the curve from labels and prediction scores.

ROC AUC Calculator calculate binary receiver-operating-characteristic area under the curve from labels and prediction scores.

Use this to evaluate how well a score ranks positive examples above negative examples across every classification threshold.

When to use ROC AUC Calculator

Use this to evaluate how well a score ranks positive examples above negative examples across every classification threshold.

How it works

Scores are sorted descending, equal-score items advance as one threshold group, and trapezoidal integration accumulates true-positive rate against false-positive rate.^[1]

Interpreting the result

Keep the input units, direction conventions, encoding, and ordering rules visible when reusing the result. Preserve the full returned value for downstream calculations and round only for presentation.

Assumptions and limitations

AUC measures ranking, not calibration or a chosen operating threshold. Both classes are required, sample weighting is not supported, and tied scores contribute partial credit.

References

An introduction to ROC analysis — Pattern Recognition Letters