Exam AWS Certified Machine Learning - Specialty topic 1 question 272 discussion

The key here is before training the model. If it is before training then we can do that using SageMaker Data Wrangler. After training and model is deployed for inference, we can use model monitor

A and C are correct. https://docs.aws.amazon.com/sagemaker/latest/dg/clarify-data-bias-reports-ui.html

A. YES - SageMaker Clarify can be used to detect bias during prep (https://aws.amazon.com/sagemaker/clarify/) B. NO - SageMaker Ground Truth is not used in the solution C. NO - drift report to during inference D. YES - Configure SageMaker Data Wrangler to generate a bias report. E. NO - SageMaker Experiments is to compare model outputs

This combination meets all the requirements with the least operational overhead. You can use SageMaker Data Wrangler to ingest and clean your data in Amazon SageMaker Studio without writing any code. You can also use SageMaker Clarify to automatically detect potential bias in your data using predefined or custom metrics. You can then configure SageMaker Data Wrangler to generate a bias report that shows the results of the bias analysis in a visual and interactive way2.

A: AWS Clarify used to generate Bias report. D: AWS Data Wrangler to generate Bias report

AD: https://docs.aws.amazon.com/sagemaker/latest/dg/clarify-data-bias-reports-ui.html

I think it's A+E

AC is correct. A: AWS Clarify used to generate Bias report. C: AWS Data Wrangler to generate Bias report and is operationally efficient. https://catalog.us-east-1.prod.workshops.aws/workshops/1e224d5a-4273-444a-acec-28d44a5bfb28/en-US/data-preparation/amazon-sagemaker/data-wrangler