Using AWS Services for Metric Tracking and Monitoring
Question
You are working on a Linear Learner algorithm-based model to predict the quarterly sales for each region of your company's global sales force.
The model needs to use data from your sales team's past sales performance, such as the quantity of products sold, revenue generated, expenses incurred, sales force size, etc. You and your team are in the process of training the model based on the SageMaker built-in Linear Learner algorithm.
You want to track and monitor metrics, such as test objective loss and test precision, as the model trains.
Which AWS service(s) would you use to track and monitor these metrics? (Select THREE)
Answers
Explanations
Click on the arrows to vote for the correct answer
A. B. C. D. E. F.Answers: B, D, E,
Option A is incorrect.
You can specify the metrics you want to track using the AWS Management Console for SageMaker, not the AWS Management Dashboard for SageMaker.
Option B is correct.
To specify the metrics you want to track, you use the AWS Management Console for SageMaker or the SageMaker Python SDK APIs.
Option C is incorrect.To specify the metrics you want to track, you use the AWS Management Console for SageMaker or the SageMaker Python SDK APIs, not the SageMaker Javascript SDK APIs.
Option D is correct.
To specify the metrics you want to track, you use the AWS Management Console for SageMaker or the SageMaker Python SDK APIs.
Option E is correct.
Once the model training starts, SageMaker streams the metrics you specified to CloudWatch, where you can visualize the time-series curves of your metrics.
Option F is incorrect.
You can visualize your metrics either via the CloudWatch console or the SageMaker Python SDK APIs, not the SageMaker Javascript SDK APIs.
Reference:
Please see the AWS Machine Learning Blog titled Easily monitor and visualize metrics while training models on Amazon SageMaker.
To track and monitor metrics for the Linear Learner algorithm-based model, the following AWS services can be used:
A. Specify the metrics you want to track using the AWS Management Dashboard for SageMaker. B. Specify the metrics you want to track using the AWS Management Console for SageMaker. D. Specify the metrics you want to track using the SageMaker Python SDK APIs. E. Use the CloudWatch console for visualizing time-series curves of your metrics.
The AWS Management Dashboard for SageMaker allows you to track the performance of the models that you have trained using SageMaker. You can specify the metrics you want to track and monitor them using the dashboard. The dashboard provides real-time metrics on the performance of the model, such as the training and testing loss, accuracy, precision, recall, F1 score, and other evaluation metrics. You can also use the dashboard to view logs and other information related to the training process.
The AWS Management Console for SageMaker also provides a way to specify the metrics you want to track and monitor. The console provides a user-friendly interface for setting up and monitoring the training process. You can view the training and testing loss, accuracy, precision, recall, F1 score, and other evaluation metrics in real-time using the console.
The SageMaker Python SDK APIs provide a way to programmatically specify the metrics you want to track and monitor during the training process. You can use the APIs to define the metrics you want to track and view the results in real-time.
The CloudWatch console can be used to visualize the time-series curves of the metrics that you are tracking. You can create custom dashboards and add the metrics you want to monitor to the dashboard. The dashboard provides a way to view the metrics over time and identify trends and patterns in the data.
In summary, to track and monitor metrics for the Linear Learner algorithm-based model, you can use the AWS Management Dashboard for SageMaker, AWS Management Console for SageMaker, SageMaker Python SDK APIs, and CloudWatch console. These services provide real-time metrics and visualization tools to help you monitor the training process and evaluate the performance of the model.
