CloudWatch ML Metrics Quiz

5 questions Pass: 70% +25 pts (50 with Pro)

Quiz covering Model Monitoring

What you'll be asked

  1. Which AWS service is primarily used to collect and store model performance metrics such as latency and invocation counts?
  2. When monitoring an Amazon SageMaker endpoint, which metric is most useful for identifying if your model is struggling to keep up with incoming request volume?
  3. You have configured a CloudWatch Alarm to trigger when the 99th percentile of model latency exceeds 200ms. Why is it better to use P99 rather than Average latency for this alarm?
  4. Which feature allows you to visualize multiple model performance metrics, such as Invocations and ModelLatency, on a single screen in the AWS Console?
  5. You notice that your SageMaker endpoint is triggering an 'Invocation4XXErrors' alarm. What is the most likely cause you should investigate first?

Answers and explanations come after you submit.