Question 3
UnclassifiedOn a highly imbalanced dataset, which metric pair is most informative?
Correct answer: B
Explanation
On a highly imbalanced dataset, "Precision-Recall AUC" is more informative than accuracy because it focuses on performance for the positive class and is sensitive to false positives and false negatives. "F1" combines precision and recall into one score, so the pair highlights how well the model finds rare positives without being misled by the majority class.
Why each option is right or wrong
A. Accuracy and ROC-AUC
B. Precision-Recall AUC and F1
On a highly imbalanced dataset, accuracy can be mathematically high even when the classifier misses most of the minority class, so the more diagnostic pair is the one built from the confusion-matrix rates that matter for rare positives. Precision-Recall AUC summarizes performance across all thresholds using precision = TP/(TP+FP) and recall = TP/(TP+FN), and F1 is the harmonic mean of those same two quantities, so together they directly reflect false positives and false negatives rather than the dominant negative class.
C. MSE and RMSE
D. R² and MAE