Skip to content

IAI Actuarial Core Principles · Risk Modelling and Survival Analysis · Elementary principles of machine learning

A classifier is tested on 200 policies, of which 40 are actual fraud cases. It flags 50 policies as fraud, and 30 of these are genuinely fraudulent. What are its precision and recall for fraud?

Precision is 0.60 and recall is 0.75. With 30 true positives, precision divides by the 50 flagged policies and recall divides by the 40 actual fraud cases. Swapping these denominators gives the common wrong answer of 0.75 and 0.60.

  1. APrecision 0.60, recall 0.75Correct
  2. BPrecision 0.75, recall 0.60
  3. CPrecision 0.60, recall 0.15
  4. DPrecision 0.75, recall 0.15
  5. Precision 0.80, recall 0.60

Explanation

True positives = 30. Precision = 30/50 = 0.60 (flagged cases that are fraud). Recall = 30/40 = 0.75 (actual frauds caught). Option 1 swaps the two measures; the others use wrong denominators such as 200.

Did you get it right without looking?

One question tells you little. A timed set on Elementary principles of machine learning shows your real accuracy, how long you take and where you lose marks.

Practise Elementary principles of machine learningFree diagnostic mock

More Elementary principles of machine learning questions