Google Professional Machine Learning Engineer Question 450
Select 4Google Cloud PlatformYour team is building a machine learning model for predicting customer churn using Google Cloud. To ensure compliance with internal policies, you must track both the data lineage and the model lineage. Which of the following steps should you take to enable proper lineage tracking?
- A
Use Vertex AI's Model Registry to track model versions and metadata.
- B
Enable Data Catalog to manage and track metadata for your datasets.
- C
Store your training data in BigQuery and use built-in audit logs to track data access.
- D
Manually document model training processes and data sources in a shared document.
- E
Use Vertex AI Pipelines to automatically capture lineage information for data transformations and model training.
Show answer and explanation
Correct answers: A, B, C, E
Explanation
Tracking model and data lineage is essential for ensuring compliance, reproducibility, and accountability in machine learning workflows. Google Cloud provides tools such as Vertex AI Model Registry, Data Catalog, and Vertex AI Pipelines to automate and streamline lineage tracking. Additionally, BigQuery's audit logs can help track the usage and modification of datasets. However, manual documentation is not a recommended approach as it lacks automation and is error-prone.
- A. Correct.
Vertex AI's Model Registry allows you to track and manage different model versions, including metadata such as creation date, performance metrics, and deployment history, which is critical for model lineage.
- B. Correct.
Data Catalog is a Google Cloud tool designed for metadata management, enabling you to track and search for datasets, which is essential for data lineage.
- C. Correct.
BigQuery provides built-in audit logs that track access and modifications to datasets, which contribute to ensuring data lineage.
- D. Incorrect.
While manually documenting processes may provide some level of tracking, it is not an automated, scalable, or reliable approach for ensuring lineage, and it is prone to human error.
- E. Correct.
Vertex AI Pipelines automatically capture detailed lineage information for both datasets and models during ML workflows, helping to ensure compliance and reproducibility.