A card issuer wants transactions that look unlike a customer's established spending pattern flagged while the payment is being authorised. Which workload should be used?
Show answer and explanation
Answer: B
Anomaly detection models the normal distribution of behaviour, here a cardholder's typical merchants, amounts, locations and times, and raises a signal when a new transaction deviates markedly. Because genuine fraud is rare and constantly changing, learning the boundary of normality is more practical than writing explicit rules for every fraud pattern. The result is usually a risk score that feeds a decision step, and analysts review flagged cases to limit false positives.
Why the other options are wrong
- A. Image classification labels pictures and cannot evaluate transaction sequences or spending behaviour.
- C. Sentiment analysis judges the emotional tone of text, which is unrelated to payment authorisation patterns.
- D. Speech recognition converts spoken audio into text and plays no part in scoring card transactions.
Exam tip. Unusual vs normal behaviour equals anomaly detection.