AWS Certified Machine Learning – SpecialtyMachine Learning Implementation and OperationsMedium
A media company uses a machine learning model to personalize content recommendations for its users. The model is deployed on an Amazon SageMaker endpoint. Over time, the performance of the model has degraded, and the data science team suspects that the underlying data distribution has changed, leading to 'concept drift'. To systematically address this, they want to automatically retrain the model when significant data changes are detected. Which SageMaker service can be used to monitor data quality and trigger model retraining based on drift detection?
- AAmazon SageMaker Ground Truth
- BAmazon SageMaker Pipelines
- CAmazon SageMaker Model Monitor
- DAmazon SageMaker Clarify
Show answer & explanationAnswer & explanation
Correct answer: C. Amazon SageMaker Model Monitor
Amazon SageMaker Model Monitor is specifically designed to detect data and model quality issues, including concept drift, in production ML models. It can analyze incoming inference requests and compare their statistics against a baseline, triggering alerts or automated retraining workflows when significant drift is detected.
Why the other options are wrong
- A. SageMaker Ground Truth is for data labeling, not model monitoring or drift detection.
- B. SageMaker Pipelines orchestrates ML workflows but does not inherently detect drift; it would be used to build the retraining pipeline triggered by Model Monitor.
- D. SageMaker Clarify is used for bias detection and explainability, not for continuous monitoring of data drift for retraining.
SageMaker Model Monitor
A SageMaker service that continuously monitors the quality of ML models in production, detecting issues such as data drift, model drift, and data quality anomalies.
- Compares real-time inference data against a baseline.
- Can trigger alerts or automated retraining pipelines.
- Helps maintain model performance over time.
Memory trick: Model Monitor is the model's watchful eye.