You are an ML engineer at a travel company. You have been researching customers’ travel behavior for many years, and you have deployed models that…

Option A is incorrect because Cloud SQL is a relational database service that is not designed for

storing and comparing model performance statistics. It would require writing complex SQL queries to

perform the comparison, and it would not provide any visualization or analysis tools.

Option B is incorrect because Vertex AI does not support creating versions of models for each season

per year. Vertex AI models are versioned based on the training data and hyperparameters, not on

external factors such as seasonality or holidays. Moreover, the Evaluate tab of the Vertex AI UI only

shows the performance metrics of a single model version, not across multiple versions.

Option C is incorrect because Kubeflow is a different platform than Vertex AI, and it does not

integrate well with Vertex AI Pipelines. Kubeflow experiments are used to group pipeline runs that

share a common goal or objective, not to compare performance statistics across different seasons or

years. Kubeflow UI does not provide any tools to compare the results across the experiments, and it

would require switching between different platforms to access the data.

Option D is correct because Vertex ML Metadata is a service that allows storing and tracking

metadata associated with machine learning workflows, such as models, datasets, metrics, and

events. Events are user-defined labels that can be used to group or slice the metadata for analysis. By

using seasons and years as events, you can easily store and compare the performance statistics of

each version of your models across different time periods. Vertex ML Metadata also provides tools to

visualize and analyze the metadata, such as the ML Metadata Explorer and the What-If Tool.