🟢 Repeat: Classification Metrics (Can You Still Explain?)

Fill in the blanks without looking:

1. Precision = TP / (TP + ___)
2. Recall = TP / (TP + ___)
3. Use precision when ___ are costly.
4. Use recall when ___ are costly.
5. For imbalanced data, use ___ instead of ROC-AUC.

Answers: 1. FP 2. FN 3. False positives (flagging innocent things) 4. False negatives (missing real positives) 5. PR-AUC (Precision-Recall AUC)