Table 1.

Metric validation. The advantages and disadvantages of receiver operating characteristics (ROC) and precision–recall (PR) evaluation metrics.

Metric	Advantages	Disadvantages
ROC	1. Simple graphical representation and exact measure of the accuracy of a test. 2. Performs equally well on both classes in balanced datasets. 3. The AUC is used as a simple numeric rating of diagnostic test accuracy.	1. Actual decision thresholds are usually not displayed. 2. As the sample size decreases, the plot becomes irregular. 3. Not considered a good indicator for early enrichment of true active samples.
PR	1. Points out the efficiency of the model. 2. Shows how much the data are biased towards one class. 3. Helps understand whether the model is performing well in imbalanced datasets.	1. It does not deal with all the cells of the confusion matrix. True negatives are never considered. 2. Focuses only on positive class. 3. Only suited for binary classification.