auc¶
-
getml.models.scores.
auc
= 'auc'¶ Area under the curve - refers to the area under the receiver operating characteristic (ROC) curve.
Used for classification problems.
When handling a classification problem, the ROC curve maps the relationship between two conflicting goals:
On the hand, we want a high true positive rate. The true positive rate, sometimes referred to as recall, measures the share of true positive predictions over all positives:
In other words, we want our classification algorithm to “catch” as many positives as possible.
On the other hand, we also want a low false positive rate (FPR). The false positive rate measures the share of false positives over all negatives.
In other words, we want as few “false alarms” as possible.
However, unless we have a perfect classifier, these two goals conflict with each other.
The ROC curve maps the TPR against the FPR. We now measure the area under said curve (AUC). A higher AUC implies that the trade-off between TPR and FPR is more benefitial. A perfect model would have an AUC of 1. An AUC of 0.5 implies that the model has no predictive value.