Classification evaluation
ROC curve & AUC calculator
Paste binary labels and model scores to calculate the ROC curve points and the area under the ROC curve. The calculator treats higher scores as stronger evidence for the positive class.
Free · no sign-in · threshold-aware
1. Enter labels and scores
Use 1 for positive and 0 for negative
Scores or probabilities in the same order
Try an example:
Metrics update as you type.
ROC-AUC0
Ranking discrimination
Quick read: enter labels and scores to see the ranking result.
ROC points by threshold
For each unique score, the calculator treats scores at or above that threshold as positive. FPR is the false positive rate; TPR is the true positive rate, also called recall.
| Threshold | TP | FP | FPR | TPR |
|---|
How ROC-AUC is calculated
TPR = TP / (TP + FN)
FPR = FP / (FP + TN)
AUC = Σ(width between ROC points × average height)
Read the ROC curve and AUC guide for a worked calculation, threshold trade-offs, and when precision–recall analysis is more informative. For direct confusion-matrix counts, use the confusion matrix calculator.