Google Professional Machine Learning Engineer
Get started today
Ultimate access to all questions.
You work as a machine learning engineer and have developed a custom model using Vertex AI to predict your application's user churn rate. For maintaining model performance, you employ Vertex AI Model Monitoring for skew detection. The training data, stored in BigQuery, contains two sets of features: demographic and behavioral. After some analysis, you discover that two separate models trained on each feature set individually perform better than the original combined model. Given this, you need to configure a new model monitoring pipeline that can split traffic among the two models. It is crucial that both models adhere to the same prediction-sampling rate and monitoring frequency while minimizing the management effort required. What should you do?
You work as a machine learning engineer and have developed a custom model using Vertex AI to predict your application's user churn rate. For maintaining model performance, you employ Vertex AI Model Monitoring for skew detection. The training data, stored in BigQuery, contains two sets of features: demographic and behavioral. After some analysis, you discover that two separate models trained on each feature set individually perform better than the original combined model. Given this, you need to configure a new model monitoring pipeline that can split traffic among the two models. It is crucial that both models adhere to the same prediction-sampling rate and monitoring frequency while minimizing the management effort required. What should you do?
Explanation:
The correct answer is B. Deploying both models to the same endpoint and submitting a Vertex AI Model Monitoring job with a monitoring-config-from-file parameter that accounts for the model IDs and feature selections minimizes the management effort. This approach keeps the training dataset as is, avoiding unnecessary complexities of splitting data or managing multiple endpoints. The monitoring-config-from-file parameter enables you to specify configurations independently for each model within the same monitoring job, ensuring efficient and granular monitoring without additional overhead.