If a model has high recall but low precision, which metric best captures the trade-off between them?

If a model has high recall but low precision, which metric best captures the trade-off between them?

• A. F-score
• B. Accuracy
• C. Precision
• D. Recall

Answer: (A)

Balancing recall and precision requires a metric that increases only when both improve. Recall shows how many actual positives you identify, while precision shows how many of your positive predictions are correct. When a model has high recall but low precision, you catch most positives but also produce many false positives, so a single measure should reflect both aspects. The F-score, specifically the F1 score, is the harmonic mean of precision and recall, so it remains high only when both are high. This makes it the best way to capture the trade-off between identifying positives and keeping predictions accurate. Other measures can be misleading in isolation—accuracy can be distorted by class imbalance, and evaluating precision or recall alone ignores the other side of the trade. If you needed to prioritize one side, you could use a weighted variant (F-beta), but for a balanced trade-off the F-score is the standard choice.

Balancing recall and precision requires a metric that increases only when both improve. Recall shows how many actual positives you identify, while precision shows how many of your positive predictions are correct. When a model has high recall but low precision, you catch most positives but also produce many false positives, so a single measure should reflect both aspects. The F-score, specifically the F1 score, is the harmonic mean of precision and recall, so it remains high only when both are high. This makes it the best way to capture the trade-off between identifying positives and keeping predictions accurate. Other measures can be misleading in isolation—accuracy can be distorted by class imbalance, and evaluating precision or recall alone ignores the other side of the trade. If you needed to prioritize one side, you could use a weighted variant (F-beta), but for a balanced trade-off the F-score is the standard choice.