Performance Evaluation

To evaluate the predictive performance of the classifier models, we calculated 6 representative performance evaluation measures, including sensitivity, specificity, positive predictive value (PPV), negative predictive value, positive likelihood ratio, and negative likelihood ratio. The ROC-AUC and the area under the precision recall curve (PR-AUC) were also computed. The recall corresponds to the sensitivity, and the precision corresponds to the PPV. The PR curve is often used along with the ROC curve to assess the model performance, especially in imbalanced data sets.