Question 29
Content Domain 4: Machine Learning Implementation and OperationsIn an ML operations workflow, what is the primary purpose of a retraining pipeline?
Correct answer: B
Explanation
A retraining pipeline is used to update an existing machine learning model by running the steps needed to train it again on data. It is specifically concerned with model retraining rather than unrelated deployment or monitoring tasks. — Source material: "Retrain pipelines. Key Terms: retraining pipelines"
Why each option is right or wrong
A. To continuously monitor infrastructure costs after a model is deployed
A retraining pipeline focuses on training workflow steps, not post-deployment cost monitoring.
B. To run the process that trains an existing model again using data
The source material identifies the topic as "Retrain pipelines" and the key term as "retraining pipelines," which directly supports that the pipeline's role is to retrain a model by running training again.
C. To route online prediction requests from applications to a serving endpoint
Serving prediction requests is an inference or deployment function, not a retraining function.
D. To create dashboards that summarize model performance metrics for stakeholders
Performance dashboards support reporting or monitoring, not the act of retraining the model.