https://doi.org/10.1351/goldbook.11502
A plot that shows the result of a test of the performance of a binary classifier by plotting the fraction of true positives identified versus the fraction of false positives as the discrimination threshold is varied.
Notes:
- ROC curves are frequently used to compare different docking/scoring combinations or virtual screening protocols by the retrieval results from seeding a database of inactive molecules with known actives.
- The area under the ROC curve (AUC) provides a measure of the superiority of the model over random predictions: If the value for the AUC for a ROC curve has the value of \(\pu{0.9\!-\! 1.0}\), the fit is considered excellent, whereas an AUC of \(\pu{0.5}\) suggests that the algorithm has no discriminatory power.
- This is an example of using area under the curve as a measure of model performance.