When prioritize precision vs recall

Context

You are part of a product team, and you are building (or evaluating) a binary classifier whose output triggers an action—for example, showing a warning, blocking content, routing to manual review, sending a notification, or prioritizing leads.

Question

1. Explain the business meaning of precision and recall in this context.
2. In what kinds of business scenarios would you favor precision over recall? Give 2–3 concrete examples.
3. In what kinds of business scenarios would you favor recall over precision? Give 2–3 concrete examples.
4. Describe how you would select an operating threshold (or decision rule) in practice, including:
   • the cost/benefit tradeoff and asymmetric error costs
   • capacity constraints (e.g., manual review bandwidth)
   • base-rate / class imbalance considerations
   • what metrics/curves you would look at (e.g., PR curve)
   • any guardrail metrics you would monitor after launch

Overview: This question tests a data scientist's grasp of binary classification metrics (precision and recall), cost-sensitive decisions, and choosing operating thresholds for production models in the machine learning domain.

Read the full data science interview experience this question came from.