Databricks Machine Learning Associate Question 630
Single answerA company wants to deploy a machine learning model to recommend personalized products to users. Users expect recommendations to update immediately after they interact with the platform (e.g., clicking or purchasing products). Which model serving approach is the most appropriate for this scenario?
- A
Batch serving, as it processes large volumes of data periodically and is cost-effective.
- B
Realtime serving, as it provides predictions with low latency based on user interactions.
- C
Streaming serving, as it processes data in micro-batches but does not require immediate predictions.
- D
Batch serving with manual refresh, as it allows manual updates to predictions when needed.
Show answer and explanation
Correct answer: B
Explanation
Realtime serving is the most appropriate solution for scenarios requiring immediate predictions or updates, such as personalized recommendations that change dynamically based on user interactions. Batch serving and streaming serving do not meet the low-latency requirements of this use case, and manual refreshes are impractical in production environments.
- A. Incorrect.
Batch serving is not suitable for this scenario because it processes data periodically, which introduces delays and does not meet the requirement for immediate updates after user interactions.
- B. Correct.
Realtime serving is the correct choice as it provides low-latency predictions, enabling the system to respond immediately to user clicks or purchases, meeting the need for instant updates.
- C. Incorrect.
Streaming serving is designed for continuous data ingestion and processing but typically operates in micro-batches, which may not provide the real-time responsiveness required for instant updates.
- D. Incorrect.
Batch serving with manual refresh is inefficient and impractical for scenarios requiring immediate updates, as it involves human intervention and does not provide real-time predictions.