association
ROC analysis
Evaluate how a numeric score separates a binary outcome across thresholds.
Requirements
- binary reference outcome
- numeric score or predicted probability
Avoid when
- uncalibrated class labels only
- clinical utility without threshold consequences
Formula
TPR = TP/(TP+FN); FPR = FP/(FP+TN); AUC = ∫TPR dFPR
Interpretation
AUC summarizes discrimination; threshold rows show sensitivity and specificity tradeoffs.
Worked example
Evaluate how well a biomarker score separates cases from controls across thresholds.
Open example datasets