Google Professional Machine Learning Engineer Question 243
Select 2Google Cloud PlatformYou are tasked with building a machine learning pipeline on Google Cloud for a retail company. The pipeline needs to ingest customer transaction data in CSV format, product catalog data in JSON format, and customer profile images stored in a Cloud Storage bucket. Which combination of services and tools would be most appropriate for ingesting these various file types efficiently into your pipeline?
- A
Cloud Storage for storing the files and Dataflow for processing the ingestion
- B
BigQuery for directly uploading and querying all file types
- C
Pub/Sub for streaming the files and Cloud Data Fusion for processing
- D
Cloud Storage for storing files, Cloud Functions for triggering ingestion, and Dataflow for processing
- E
Dataproc for ingesting and transforming all file types
Show answer and explanation
Correct answers: A, D
Explanation
Ingesting various file types such as CSV, JSON, and images requires a combination of storage, event-triggering, and processing tools. Cloud Storage is a natural choice for storing all file types, while Dataflow is well-suited for processing structured and unstructured data. Cloud Functions can act as an event-driven service to trigger ingestion workflows when new files arrive. BigQuery and Dataproc may be used in later stages of the pipeline but are not optimal for ingestion of diverse file types.
- A. Correct.
Cloud Storage is a scalable and cost-effective choice for storing various file types, and Dataflow is ideal for processing and transforming structured and unstructured data during ingestion.
- B. Incorrect.
BigQuery is not suitable for directly uploading and querying all file types (e.g., images). It is better suited for querying structured data, such as CSV or JSON, that has already been ingested into the pipeline.
- C. Incorrect.
Pub/Sub is primarily used for streaming data in real-time, but it is not suitable for batch ingestion of static files like images or CSV/JSON files.
- D. Correct.
This is a valid approach. Cloud Storage can store the files, Cloud Functions can trigger ingestion tasks when new files are uploaded, and Dataflow can process and transform the data.
- E. Incorrect.
Dataproc is better suited for big data processing using Hadoop/Spark, and it is not the most efficient option for ingesting a variety of file types like images and structured data.