Google Professional Machine Learning Engineer Question 301
Select 2Google Cloud PlatformYou are building a machine learning system for a retail company that predicts daily sales for each store location. The company requires real-time predictions for its online store to update product availability and batch predictions for its physical stores' inventory planning every night. Which combination of tools or services would best meet these requirements?
- A
Use Vertex AI for online inference and Dataflow for batch predictions.
- B
Use BigQuery ML for online inference and Vertex AI Batch Prediction for batch predictions.
- C
Use Vertex AI for online inference and Vertex AI Batch Prediction for batch predictions.
- D
Use Dataproc for online inference and Dataflow for batch predictions.
- E
Use BigQuery ML for online inference and Dataflow for batch predictions.
Show answer and explanation
Correct answers: A, C
Explanation
For this scenario, real-time predictions for the online store require a tool that supports online inference, such as Vertex AI. Batch predictions for the physical stores' inventory planning can be handled by either Vertex AI Batch Prediction or Dataflow, as both are capable of handling batch processing efficiently. Vertex AI offers a unified platform for both online and batch predictions, making it an ideal choice in this case.
- A. Correct.
Correct. Vertex AI supports online inference through deployed models and can be used for real-time predictions, while Dataflow is a great choice for processing large-scale data in batch mode.
- B. Incorrect.
Incorrect. BigQuery ML is primarily designed for training and running SQL-based machine learning models and is not optimized for real-time online inference.
- C. Correct.
Correct. Vertex AI supports both online inference through deployed models and batch predictions, making it a suitable choice for this scenario.
- D. Incorrect.
Incorrect. Dataproc is a managed Hadoop/Spark service and is not typically used for online inference. It is better suited for large-scale data processing jobs.
- E. Incorrect.
Incorrect. BigQuery ML is not designed for real-time predictions, and Dataflow is not typically used for online inference either.