mediumMultiple Select

Generative AI Leader Practice Question: A data scientist wants to train a model using…

A data scientist wants to train a model using BigQuery ML. Which two statements are true about BigQuery ML? (Choose two.)

⚠ Common exam trap

Generative AI Leader often tests the misconception that BigQuery ML requires data export or external training infrastructure — the exam expects recognition that BigQuery ML trains models in-database using SQL and supports both supervised and unsupervised learning.

Answer choices

Why each option matters

Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.

Correct answer & explanation

✓

Models can be trained using SQL directly on BigQuery data

Option D is correct because BigQuery ML's core design lets you create and train models directly on data stored in BigQuery using standard SQL statements such as CREATE MODEL, without moving or exporting the data. Option E is correct because BigQuery ML supports supervised learning (for example, linear regression, logistic regression, and boosted trees) as well as unsupervised learning (for example, k-means clustering and matrix factorization). Option A is wrong because BigQuery ML performs training inside BigQuery itself and does not require a separate Vertex AI training cluster. Option B is wrong because BigQuery ML supports many model types beyond linear regression, including logistic regression, k-means, and deep neural networks. Option C is wrong because training happens directly on BigQuery tables, so exporting data to Cloud Storage first is unnecessary.

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

✗
It requires a separate Vertex AI training cluster
Why it's wrong here
BigQuery ML trains models inside BigQuery using SQL, with no separate Vertex AI training cluster required. It is tempting because Vertex AI is Google Cloud's dedicated ML platform, but that is the choice when custom training code, pipelines or serving infrastructure are needed instead.
✗
It supports only linear regression models
Why it's wrong here
BigQuery ML supports linear regression alongside logistic regression, k-means, matrix factorisation, boosted trees, deep neural networks and imported TensorFlow models, so restricting it to linear regression misstates its model catalogue. Linear regression alone is genuinely available, which makes the option tempting for simple forecasting tasks where only that algorithm is needed.
✗
Data must be exported to Cloud Storage before training
Why it's wrong here
BigQuery ML trains directly on BigQuery tables using SQL, so no export to Cloud Storage is required; the data stays in place. Exporting to Cloud Storage is genuinely needed when training with Vertex AI custom containers or other external frameworks, which makes the step sound familiar but unnecessary here.
✓
Models can be trained using SQL directly on BigQuery data
Why this is correct
BigQuery ML lets you create and train models using SQL queries directly against data already stored in BigQuery, avoiding data export or separate ML tooling. This satisfies the stem's requirement to train within BigQuery using familiar SQL syntax.
✓
It supports both supervised and unsupervised learning
Why this is correct
BigQuery ML supports supervised algorithms such as linear regression and logistic regression alongside unsupervised ones including k-means clustering, so both paradigms are covered. This breadth is the factual basis for the statement, distinguishing BigQuery ML from tools limited to a single learning type.

Go deeper

Generative AI Leader Practice Question: A data scientist wants to train a model using…

Why each option matters

Option-by-option breakdown

Related to this question