The receiver operating characteristic, or ROC, curve is a popular plot for simultaneously displaying the tradeoff between the true positive rate and the false positive rate for a binary classifier at different classification thresholds. The ROC curve dates back to World War II, when it was used initially to analyze radar signals and later in signal detection.
The overall performance of a classifier, summarized over all possible classification thresholds, is given by the area under the ROC curve, or AUC. An ideal ROC curve will hug the top left corner, indicating a high true positive rate and a low false positive rate; the larger the AUC the better the classifier. AUC represents the probability that the model ranks a random positive case higher than a random negative case. AUC ranges from 0 to 1, where an AUC of 0 indicates that the model got all predictions wrong while an AUC of 1 indicates that the model got all predictions right.
As an important classification performance measure, ROC curves are useful for comparing different classifiers because they consider all possible thresholds. Additionally, the initial slope of the ROC curve indicates how quickly performance degrades down the ranked list of predictions.
An ROC curve can be plotted for every binary classification trained on the C3 AI Platform. The AUC can also be used as an evaluation criterion for an MLPipeline and can be used for hyperparameter optimization.