Skip to content

ACCA Strategic Professional · Advanced Performance Management · Technology and information systems

Kovar Bank uses a machine learning model to flag suspicious card transactions. In a test on 10,000 transactions, 200 were genuinely fraudulent. The model flagged 250 transactions, of which 150 were genuinely fraudulent. The performance management team wants a measure of how many of the actual frauds the model detected. Which measure and value is correct?

Recall of 75% is correct. Recall measures the proportion of actual frauds that the model caught, which is 150 true positives divided by 200 genuine frauds. The 60% figure is precision, which shows how many flagged transactions were really fraudulent, not how many frauds were detected.

  1. APrecision of 60%, being 150 divided by 250
  2. BRecall of 75%, being 150 divided by 200Correct
  3. CRecall of 60%, being 150 divided by 250
  4. DAccuracy of 75%, being 150 divided by 200

Explanation

The share of actual frauds detected is recall (sensitivity) = true positives 150 / actual frauds 200 = 75%. Precision is 150/250 = 60%, which measures how many flagged items were truly fraud, so it answers a different question. Accuracy would use all 10,000 transactions.

Did you get it right without looking?

One question tells you little. A timed set on Technology and information systems shows your real accuracy, how long you take and where you lose marks.

Practise Technology and information systemsFree diagnostic mock

More Technology and information systems questions