Exam AWS Certified Machine Learning - Specialty topic 1 question 155 discussion

I would go with B: https://docs.aws.amazon.com/sagemaker/latest/dg/model-monitor-create-baseline.html

The baseline job computes baseline statistics and constraints for the new training set. By using this updated baseline, Model Monitor can better detect any drift or violations in the production traffic.

B. Run the Model Monitor baseline job again on the new training set: This is a key step after retraining the model. Since the model has been retrained with a new dataset, the baseline against which its predictions are compared should also be updated. Running the baseline job again on the new training set and configuring Model Monitor to use this new baseline will ensure that the monitoring is relevant to the current state of the model and the data it's processing. D. Retrain the model again with a combination of the original and new training sets: While retraining the model can be a good approach in some scenarios, there's no indication in this case that the issue lies with the model's performance itself. The issue seems to be with the Model Monitor's baseline not aligning with the current model.

https://docs.aws.amazon.com/sagemaker/latest/dg/model-monitor-interpreting-violations.html

running the Model Monitor baseline job again on the new training set and configuring Model Monitor to use the new baseline, is the most appropriate step to resolve the violations and ensure the SageMaker endpoint's performance is in line with expectations.

Running the Model Monitor baseline job again with the new training set and configuring Model Monitor to use the new baseline is a valid option to resolve the violations. By running the baseline job with the new training set, a new baseline is created, which can be used to compare with the new data to detect any drifts in the data distribution. Then, the updated baseline can be set as the new baseline for monitoring the endpoint. So, option B is also a valid solution to resolve the violations.