Skip to content
Quantoros

How do you choose between precision and recall for a given business problem?

Classic MLLevel: IntermediateClassical machine learning

How do you choose between precision and recall for a given business problem?

Hints
  • Write TP, FP, TN and FN for a small example.
  • Separate ranking quality, threshold selection and probability calibration. Which error costs more?

Answer

It depends on the relative cost of false positives vs. false negatives. Prioritize recall when missing a positive is costly (e.g., cancer screening, fraud detection where fraud slipping through is expensive). Prioritize precision when acting on a false positive is costly (e.g., flagging a legitimate transaction as fraud and blocking a customer, or spam filters wrongly blocking important email). When both matter, use F1-score or optimize a threshold against a business cost function directly rather than a symmetric metric.

Source and licence: Ishan Dutta — Awesome Data Science Interview Q&A · MIT

Related questions