An ML engineer is evaluating several ML models and must choose one model to use in production. The cost of false negative predictions by the models is much higher than the cost of false positive predi

Question

Options

• ALow precision
• BHigh precision
• CLow recall
• DHigh recall

• A
  6% (2)
• B
  3% (1)
• C
  16% (5)
• D
  75% (24)

Explanation

D is correct because when false negatives carry a high cost, you need a model that minimizes missed positives - and recall directly measures this: Recall = TP / (TP + FN). A high recall means the model catches most true positives, keeping false negatives low.

Why the distractors are wrong:

• A (Low precision) is actively undesirable - low precision means many false positives, which adds unnecessary cost even if it's the lesser concern here.
• B (High precision) optimizes for reducing false positives, which is the lower-cost error in this scenario - wrong priority.
• C (Low recall) is the opposite of what you want; low recall means many false negatives slipping through, directly worsening the high-cost outcome.

Memory tip: Think of a cancer screening test - missing a real cancer case (false negative) is far worse than a false alarm (false positive). In such life-or-death scenarios, you want high recall to catch every possible case. Whenever the problem says "false negatives are costly," map it to: high recall is the goal.

No community discussion yet for this question.