MLS-C01 Question 177
Single answerYou are working for a ride-sharing company that wants to improve its ETA (Estimated Time of Arrival) predictions for customer rides. You need to decide whether to use a batch inference or real-time inference approach for this use case. Which option best describes the most suitable approach and why?
- A
Use batch inference because it can precompute ETAs for all rides and update predictions periodically.
- B
Use real-time inference because ETAs need to be updated dynamically based on real-time traffic and ride conditions.
- C
Use batch inference because it reduces infrastructure costs compared to real-time inference for time-sensitive predictions.
- D
Use real-time inference because it provides offline precomputed ETAs that can be accessed quickly when needed.
Show answer and explanation
Correct answer: B
Explanation
ETA predictions for a ride-sharing service require real-time inference because they depend on factors such as current traffic conditions, weather, and driver location, which can change rapidly. Batch inference is better suited for use cases where predictions do not need to be updated frequently or are not highly time-sensitive.
- A. Incorrect.
Batch inference is unsuitable because ETAs are dynamic and require updates based on real-time conditions like traffic, which batch processing cannot handle efficiently.
- B. Correct.
This is the correct answer as ETAs for rides are highly dynamic and rely on real-time factors such as traffic, road closures, and current driver location. Real-time inference can provide up-to-date predictions at the moment they are needed.
- C. Incorrect.
While batch inference can reduce infrastructure costs, it is not suitable for time-sensitive, dynamically changing predictions like ETAs.
- D. Incorrect.
This is incorrect because real-time inference is not about precomputing data offline; instead, it generates predictions on the fly in response to live inputs.