Data Engineering for ML Quiz - Competitive Exams GCP_ML_Engineer

Free practice quiz: Data Engineering for ML (Competitive Exams GCP_ML_Engineer). 50 multiple-choice questions with answers and explanations. Play online, no sign-up needed.

This Competitive Exams GCP_ML_Engineer practice quiz has 50 multiple-choice questions on Data Engineering for ML. Try the 5 sample questions below (tap an option to check yourself), then play the full quiz online. No sign-up needed.

Sample questions

1. You need to efficiently preprocess a 10 TB dataset in Cloud Storage before training a model. Which service is best for scalable, distributed data processing with built-in ML-friendly transformations?

   1. Cloud SQL
   2. Cloud Dataflow
   3. Cloud Pub/Sub
   4. Cloud Scheduler
   Show answer

   Answer: B. Cloud Dataflow

2. Which BigQuery feature is best for creating a time-series feature table where queries can efficiently filter by date?

   1. Date-partitioned tables
   2. External tables
   3. Clustered tables
   4. Materialized views
   Show answer

   Answer: A. Date-partitioned tables

3. You need to deduplicate records in a large dataset using Apache Beam. Which transform is most appropriate?

   1. GroupByKey
   2. Flatten
   3. Distinct
   4. Combine
   Show answer

   Answer: C. Distinct

4. In Apache Beam, which transform would you use to apply a machine learning model to each element in a PCollection for online predictions?

   1. Combine
   2. GroupByKey
   3. ParDo
   4. Flatten
   Show answer

   Answer: C. ParDo

5. Which BigQuery feature allows you to define and enforce data quality rules on feature tables, such as checking for nulls or range constraints?

   1. BigQuery Data Quality
   2. BigQuery ML
   3. BigQuery Omni
   4. BigQuery BI Engine
   Show answer

   Answer: A. BigQuery Data Quality

More questions in this quiz

Answers and explanations are in the full quiz.

1. Your team needs to convert raw CSV files into TFRecord format for training. Which tool is most appropriate for this preprocessing step on Google Cloud?
2. Which feature of BigQuery allows you to run SQL queries directly against data stored in Cloud Storage without loading it into BigQuery?
3. You want to build a real-time feature pipeline for an online prediction model. Which Google Cloud service should you use to ingest and process streaming data with low latency?
4. Which data format is most efficient for storing large-scale training data and is natively supported by TensorFlow?
5. When preparing training data, you need to handle missing values and normalize numerical features consistently between training and serving. Which component helps avoid training-serving skew?
6. Which Google Cloud service is a fully managed, serverless, petabyte-scale data warehouse optimized for analytical queries and often used for feature engineering?
7. You need to process a large dataset with Apache Spark on Google Cloud without managing infrastructure. Which service should you use?
8. Which Dataflow feature allows you to reuse the same transformation logic for both batch and streaming pipelines?
9. You want to create a feature store that serves features with low latency for online predictions and also provides point-in-time correct features for training. Which Google Cloud service is designed for this?
10. You need to join a large BigQuery table with a small lookup table to enrich features. Which join type minimizes data shuffling?
11. Which Google Cloud service can be used to orchestrate a series of data processing steps (e.g., Dataflow, BigQuery, Dataproc) with dependencies and scheduling?
12. You need to perform exploratory data analysis on a large dataset using SQL and interactive queries. Which service is most suitable?
13. Which method in Apache Beam is used to read data from a source in a pipeline?
14. When using Dataflow, what is the purpose of a PCollection?
15. You need to handle late-arriving data in a streaming pipeline. Which Apache Beam concept allows you to define how late data is processed?
16. Which Google Cloud service is best for storing and serving large-scale structured data for low-latency online feature retrieval?
17. You want to use BigQuery ML to train a model directly on data in BigQuery. Which statement is true?
18. Which Dataflow runner is used for executing pipelines on Google Cloud's managed service?
19. Which file format is columnar and optimized for analytical queries, often used with BigQuery and Spark?
20. You need to create a training dataset with a reproducible split into train/validation/test sets. Which approach is best?
21. Which Google Cloud service provides a managed Apache Hadoop and Spark environment?
22. You want to monitor the freshness of your feature data and trigger retraining when data drifts. Which Google Cloud service can help detect anomalies in data distribution?
23. Your ML pipeline reads features from BigQuery, but the training job in Vertex AI cannot access the dataset because the service account lacks permissions. Which IAM role should you grant to the Vertex AI custom training service account to allow read access to BigQuery data?
24. You are designing a data pipeline to ingest streaming clickstream events into BigQuery with minimal latency and exactly-once processing semantics. Which combination of Google Cloud services should you use?
25. You need to store ML feature data in a format that supports efficient random access by key and is optimized for low-latency online serving. Which Google Cloud storage option is most suitable?
26. Your team wants to version control both data and code for ML experiments to ensure reproducibility. Which Google Cloud service should you use to track data lineage and dataset versions?
27. You need to create a Dataflow pipeline that reads from a Kafka cluster outside Google Cloud and writes to BigQuery. Which I/O connector should you use to read from Kafka?
28. You are using BigQuery to prepare features and notice that a query scans the entire table even though you only need the last 7 days of data. What should you do to reduce query cost and improve performance?
29. You need to ingest data from multiple sources (Cloud Storage, Pub/Sub, and BigQuery) into a single Dataflow pipeline. Which Apache Beam transform allows you to merge multiple PCollections of the same type?
30. Your ML model requires features computed with a sliding window over streaming data. Which Apache Beam windowing strategy should you use?
31. When using Dataflow, what is the role of a DoFn?
32. You are building a feature engineering pipeline that requires joining two large datasets (terabytes each) in BigQuery. Which join strategy is most efficient?
33. You need to ensure that your BigQuery ML model can be exported for online serving with low latency. Which export format is supported by BigQuery ML for TensorFlow models?
34. Which Google Cloud service provides a fully managed, serverless, and scalable way to run Apache Spark jobs that can read from BigQuery and Cloud Storage?
35. You want to create a feature store that supports both online and offline serving from a single source of truth. Which Vertex AI service should you use?
36. In a Dataflow streaming pipeline, you need to handle duplicate records that may arrive due to at-least-once delivery. Which Apache Beam transform is designed to remove duplicates based on a key within a window?
37. You are using Cloud Composer to orchestrate a daily ML training pipeline. Which component of Composer is responsible for scheduling and monitoring workflows?
38. You need to process a large dataset stored in Cloud Storage using SQL-like transformations without moving the data. Which Google Cloud service should you use?
39. In Apache Beam, what is the purpose of a watermark?
40. You need to create a reproducible train/validation/test split for a large dataset in BigQuery. Which SQL function is best suited for this?
41. Which Google Cloud service provides a managed Apache Hadoop and Spark environment that can be integrated with Cloud Storage and BigQuery?
42. You need to monitor the freshness of feature data in BigQuery and trigger retraining when data drifts. Which Google Cloud service should you use to detect anomalies and send alerts?
43. In a Dataflow pipeline, you want to write output to multiple BigQuery tables based on a field in each record. Which BigQueryIO feature allows this?
44. You are preparing features for a model that requires categorical variables to be one-hot encoded. Which BigQuery ML preprocessing function can automatically handle this?
45. You need to process a large dataset with Apache Spark on Google Cloud and want to minimize infrastructure management. Which deployment option should you choose?

More quizzes

• ML Problem Framing▶ Play
• Model Deployment & Monitoring▶ Play
• Vertex AI & Pipelines▶ Play

Browse all GCP_ML_Engineer quizzes

More to play