Google Professional Machine Learning Engineer Question 112
Select 3Google Cloud PlatformYou are building a machine learning solution on Google Cloud using Vertex AI. You need to manage a large dataset for training your model. The dataset resides in Cloud Storage, and you want to ensure it integrates seamlessly with Vertex AI for further processing. Which of the following steps should you take to properly manage the dataset in Vertex AI?
- A
Create a managed dataset in Vertex AI and import the data from Cloud Storage.
- B
Use the Vertex AI Feature Store to directly upload the raw dataset.
- C
Organize your data into a consistent structure within Cloud Storage before importing it into Vertex AI.
- D
Set up an AutoML dataset in Vertex AI and define the target column for training.
- E
Use BigQuery to preprocess the dataset and link it to Vertex AI.
Show answer and explanation
Correct answers: A, C, D
Explanation
To manage datasets in Vertex AI, you should first create a managed dataset and import your data from Cloud Storage. Ensuring that the data follows a consistent structure in Cloud Storage is a best practice for seamless integration. Additionally, if you are using AutoML, you need to define the target column during the dataset setup process. While tools like BigQuery and Feature Store are useful for specific use cases, they are not directly applicable to managing raw datasets for training in Vertex AI.
- A. Correct.
Correct: Creating a managed dataset in Vertex AI and importing data from Cloud Storage is a standard approach to managing datasets for training and evaluation.
- B. Incorrect.
Incorrect: Vertex AI Feature Store is used for managing and serving features, not raw datasets. It is not suitable for directly uploading a raw dataset.
- C. Correct.
Correct: Organizing your data into a consistent structure (e.g., following best practices for folder organization and file naming) is critical to ensuring smooth integration with Vertex AI.
- D. Correct.
Correct: Setting up an AutoML dataset in Vertex AI and defining the target column is necessary when using AutoML for training models. This helps Vertex AI understand the prediction target for the dataset.
- E. Incorrect.
Incorrect: While BigQuery is a powerful tool for preprocessing data, linking it directly to Vertex AI is not a primary step for dataset management. Instead, BigQuery is typically used for querying and transforming data before it is imported into Vertex AI.