Google Professional Machine Learning Engineer Question 235
Select 3Google Cloud PlatformYour team is building a machine learning pipeline to analyze customer transaction data stored in a Google Cloud Storage bucket. The data is in CSV format with millions of rows and is updated daily. The team plans to use BigQuery for querying and analyzing this data. Which of the following steps will help you efficiently set up and query the dataset in BigQuery?
- A
Set up a scheduled transfer in BigQuery to automatically load the updated data from the Cloud Storage bucket.
- B
Use BigQuery federated queries to directly query the CSV files stored in the Cloud Storage bucket without loading them into BigQuery.
- C
Manually upload the updated CSV files into BigQuery every day using the BigQuery console.
- D
Create an external table in BigQuery linked to the Cloud Storage bucket and use it to query the data.
- E
Enable BigQuery's streaming API to ingest the data from Cloud Storage in real-time.
Show answer and explanation
Correct answers: A, B, D
Explanation
BigQuery offers multiple ways to work with data stored in Cloud Storage. Using scheduled transfers or federated queries eliminates the need for manual intervention, while external tables provide a cost-effective solution for querying large datasets without data duplication. Real-time streaming APIs are not applicable here as the updates are batch-based and occur daily.
- A. Correct.
Correct: A scheduled transfer simplifies the pipeline by automating the data load process from Cloud Storage to BigQuery, ensuring up-to-date data for analysis.
- B. Correct.
Correct: BigQuery federated queries allow you to query the data in Cloud Storage directly, avoiding the need for data duplication or additional storage costs.
- C. Incorrect.
Incorrect: Manually uploading data daily is not efficient or scalable for large datasets and contradicts best practices for automation.
- D. Correct.
Correct: Creating an external table in BigQuery linked to Cloud Storage allows you to query the data without loading it into BigQuery, making this an efficient option for large datasets.
- E. Incorrect.
Incorrect: BigQuery's streaming API is designed for real-time data ingestion from sources like apps or IoT devices, not for batch data stored in Cloud Storage.