Databricks Machine Learning Associate Question 626
Select 2An e-commerce company wants to implement product recommendation models for its customers. They need to provide recommendations in near-real-time on the website, process hourly logs to analyze customer behavior trends, and use a weekly report to summarize model performance. Which model serving approaches should be used for these requirements?
- A
Use real-time serving for providing recommendations on the website.
- B
Use batch serving to process hourly logs and analyze customer behavior trends.
- C
Use streaming serving to process hourly logs and analyze customer behavior trends.
- D
Use batch serving for the weekly report summarizing model performance.
- E
Use streaming serving for the weekly report summarizing model performance.
Show answer and explanation
Correct answers: A, D
Explanation
The scenario requires different model serving approaches for different tasks. Real-time serving is ideal for low-latency use cases like providing on-demand recommendations. Batch serving is better suited for periodic tasks like weekly reports. Streaming serving is not applicable to hourly processing or weekly summarizations in this case, as it is designed for continuous event-based processing.
- A. Correct.
Real-time serving is the appropriate choice for providing recommendations on the website, as it allows predictions to be made on demand with low latency, which is critical for an interactive user experience.
- B. Incorrect.
Batch serving is not suitable for processing hourly logs because it is designed for large-scale, periodic jobs, and hourly processing would typically require more frequent updates than batch jobs can provide efficiently.
- C. Incorrect.
Streaming serving is more suitable for event-driven, continuous data processing rather than hourly updates. For hourly logs, a batch approach or a hybrid approach would typically be more efficient.
- D. Correct.
Batch serving is well-suited for generating weekly reports, as these are periodic jobs that do not require real-time or low-latency processing.
- E. Incorrect.
Streaming serving is not suitable for weekly reports, as it focuses on continuous and near-real-time processing rather than periodic summarization tasks.