Task 4.1: Monitor Model Inference MLA-C01 Practice Questions (Page 2)

Questions 6–10

1. 6foundation · easy

   Which statement best describes what monitoring model inference entails?

   • ATuning hyperparameters during the training phase.
   • BEnsuring the training data is free of bias and errors.
   • CTracking the model's performance and behavior in the production environment after deployment.
   • DEvaluating the model on a held-out test set before deployment.
   Reveal Solution Discussion Select an answer first
2. 7application · medium

   A manufacturing company has a model that predicts equipment failure. Monitoring detects concept drift, and the model's performance has dropped significantly. The team needs to decide on a remediation action. What is the most appropriate action?

   • AIncrease the endpoint's instance count to improve performance.
   • BRoll back to the previous model version that was trained on older data.
   • CRetrain the model using recent labeled data that reflects the current relationship.
   • DAdjust the model's confidence threshold to reduce false positives.
   Reveal Solution Discussion Select an answer first
4. 8application · medium

   A logistics company has a model that predicts delivery delays. Monitoring detects significant data drift in the input features. The model's performance has degraded. The team needs to take action to restore performance. What is the most appropriate first step?

   • ARoll back to a previous model version that was trained on older data.
   • BIncrease the endpoint's instance count to handle more traffic.
   • CAdjust the model's decision threshold to improve performance.
   • DRetrain the model on the most recent data that reflects the current distribution.
   Reveal Solution Discussion Select an answer first
5. 9application · medium

   A financial services company has a model that approves loan applications. The model was trained when interest rates were low. Now that interest rates have risen, the relationship between the input features (like credit score and income) and the target (loan approval) has changed. The model's performance is degrading. What should the team do to detect this type of change?

   • AUse SageMaker Model Monitor to compare the distribution of input features over time.
   • BMonitor the model's prediction accuracy against a holdout set of recent labeled data.
   • CUse SageMaker Clarify to generate a SHAP report for the model.
   • DSet up a CloudWatch alarm on the model's error rate to detect when it exceeds a threshold.
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   Which approach is commonly used to detect data drift?

   • AComparing the distribution of recent input data to the training data distribution.
   • BChecking the model's accuracy on a fixed test set.
   • CMonitoring the model's loss on the training set.
   • DReviewing the model's hyperparameter settings.
   Reveal Solution Discussion Select an answer first