Data pipeline and representativeness · Question 51 of 60
A model was trained and deployed after a successful representativeness test. What remains to be done in operation?
1 answer is correct. Hard
Show answer
Answer
- C Continuously monitor the properties of the operational data to detect data drift
Explanation
Representativeness is checked before training. After rollout, the real input data continues to be observed so that it becomes apparent when it moves away from the original training data (data drift).