Prepare and test your skills

Prepare and test your skills

Free sample question with answers and detailed explanation below. The Cloud Digital Leader practice bank has an additional 3,070 practice questions.Try 10 free realistic questions: instant score, no credit card, and no account required.
A business intelligence team stores mobile gaming user activity within BigQuery. The team wants to build a machine learning solution entirely with SQL in BigQuery ML to achieve two specific goals:
Which machine learning model type and built-in SQL function should the team use to meet these requirements?
A regression model predicts continuous numerical values (such as predicting revenue, house prices, or user lifetime spend) rather than discrete categorical labels. The ML.PREDICT function in BigQuery ML is used to generate inference results and predictions on new or unseen input data.
ML.PREDICT outputs predictions for individual rows rather than generating aggregate evaluation metrics such as precision, recall, and overall model accuracy.ML.PREDICT is used during operational scoring and batch prediction workflows, not model evaluation.ML.PREDICT does not produce evaluation summary tables containing precision or recall.This approach would be appropriate if the team needed to forecast the exact continuous dollar amount a player will spend in the game and subsequently generate batch predictions on newly loaded customer records.
While regression and classification are both common supervised learning techniques supported in BigQuery ML, regression is strictly used for continuous numeric targets.
BigQuery ML enables data analysts and developers to create, train, and execute machine learning models directly inside the data warehouse using standard SQL queries. A classification model (such as logistic regression or boosted trees) is designed to predict qualitative or categorical class labels—such as whether a gamer will churn (yes/no). The built-in ML.EVALUATE SQL function is used to calculate and display quantitative performance metrics against test data.
ML.EVALUATE calculates standard classification metrics, including precision, recall, accuracy, F1 score, and the area under the ROC curve (AUC-ROC), allowing the team to assess model quality without moving data.ML.EVALUATE automatically computes relevant evaluation metrics tailored to the specific model type (e.g., precision/recall for classification).Choosing a classification model with ML.EVALUATE aligns directly with the business objective of predicting churn and reviewing precision and recall metrics within the data warehouse.
Clustering is an unsupervised machine learning technique (such as k-means) used to group data points into distinct clusters based on feature similarities without using predefined or labeled target outcomes. The ML.PREDICT function applies the trained clustering model to assign cluster IDs to input records.
ML.PREDICT outputs cluster assignments rather than standard supervised evaluation metrics like precision and recall.The team wanted to discover natural player archetypes or segment users into behavioral groups without having a specific target label in mind.
Supervised classification requires labeled data and is evaluated with ML.EVALUATE to measure how accurately the model predicts the known class label.
ML.FEATURE_INFO is a built-in BigQuery ML inspection function used to retrieve statistical information and metadata about the input features used during model training, such as feature names, data types, category counts, and null value percentages.
ML.FEATURE_INFO does not evaluate model performance or calculate accuracy metrics.The team needed to inspect the descriptive statistics, categorical distributions, or schema properties of the input features used to train their BigQuery ML model.
To measure how accurately a model classifies test items, teams must use ML.EVALUATE, whereas feature inspection functions are primarily utilized during exploratory data analysis and pipeline debugging.
Answer them yourself and get your score the moment you finish. No account needed.
Full access includes
Keep the momentum going with these hand-picked practice scenarios
Last updated Validated