Google Professional Machine Learning Engineer Question 99
Select 2Google Cloud PlatformYou are working on a machine learning project that involves multiple data types: tabular data representing customer transactions, text data from customer reviews, and image data from product photos. Your team wants to organize the data effectively for preprocessing and training. What approach should you take to structure and store this data on Google Cloud to ensure scalability and ease of access for downstream machine learning tasks?
- A
Store tabular data in BigQuery, text data in Cloud Storage, and image data in Cloud Storage.
- B
Store all data types in a single Cloud SQL database for centralized access.
- C
Use Cloud Storage with separate buckets for text and image data, and store metadata in BigQuery for tabular data.
- D
Utilize Firestore to store both tabular and text data, and Cloud Storage for image data.
- E
Store all data types in a single BigQuery table using base64 encoding for image data and text fields.
Show answer and explanation
Correct answers: A, C
Explanation
Organizing data in Google Cloud requires choosing services that are optimized for specific data types. BigQuery is ideal for tabular data due to its scalability and advanced analytical capabilities. Cloud Storage is the preferred choice for unstructured data such as text and images, as it provides flexible storage options and integrates seamlessly with ML workflows. Separating data storage by type ensures that each dataset is stored in an optimal format, improving scalability, performance, and ease of integration with machine learning pipelines.
- A. Correct.
Correct: BigQuery is a scalable solution for tabular data and Cloud Storage is optimal for unstructured data like text and images. This approach ensures scalability and compatibility with Google Cloud's ML tools.
- B. Incorrect.
Incorrect: While Cloud SQL can store different types of data, it is not optimized for large-scale unstructured data like images and text files, making it an inefficient choice for this scenario.
- C. Correct.
Correct: Using Cloud Storage for unstructured data and BigQuery for tabular metadata is a common and scalable practice. This separation allows for efficient data access and organization.
- D. Incorrect.
Incorrect: Firestore is a NoSQL database suited for document-based data, but it is not ideal for large-scale analytical workloads or storing unstructured data like images.
- E. Incorrect.
Incorrect: Storing all data in a single BigQuery table with base64 encoding for images and text fields is inefficient, as BigQuery is not optimized for storing and processing large unstructured files like images or text directly.