QA Fundamentals · Senior · Question 50 of 76
How do you test a machine learning feature such as recommendations?
- 1Skip testing, since models can't be tested at all
- 2Check data, offline metrics, edge cases and monitor results live
- 3Expect one exact output for each input like normal code
Show the answer
Correct answer: Check data, offline metrics, edge cases and monitor results live
Explanation
ML outputs are probabilistic. Testers check data quality, metrics on held-out data, behaviour for empty or unusual users, bias, and drift after release.