A spam filter flags 100 emails and 80 of them really are spam. The inbox held 200 spam emails in total. What are precision and recall?

Options

1. Precision 0.4, recall 0.4
2. Precision 0.4, recall 0.8
3. Precision 0.8, recall 0.8
4. Precision 0.8, recall 0.4 (Correct option)

Answer

Precision 0.8, recall 0.4

Why

Precision = TP / (TP + FP) = 80 / 100 = 0.8: of what you flagged, how much was right.

Recall = TP / (TP + FN) = 80 / 200 = 0.4: of all the real spam, how much you caught.

Precision uses the flagged emails as its denominator; recall uses the real positives.

Remember

Precision: out of flagged. Recall: out of real.

Key terms

• Precision
• Recall (sensitivity)
• Confusion matrix