Databricks Machine Learning Associate Question 627
Single answerA data engineering team is tasked with deploying a machine learning model to predict fraudulent transactions. The model must provide predictions in near real-time, as the transactions are processed, to prevent fraudulent activities. Which model serving approach is the most suitable for this scenario?
- A
Batch serving
- B
Real-time serving
- C
Streaming serving
- D
Offline serving
Show answer and explanation
Correct answer: B
Explanation
Real-time serving is the most suitable approach for the given scenario because it ensures that predictions are made immediately as new data is received. This capability is crucial for tasks like fraud detection, where delays in prediction could result in significant losses. Other serving methods, such as batch or offline serving, introduce latency that is unacceptable for real-time requirements.
- A. Incorrect.
Batch serving processes data in bulk at scheduled intervals, which is not suitable for near real-time prediction needs.
- B. Correct.
Real-time serving provides predictions as new data arrives, making it the most appropriate choice for scenarios requiring immediate responses like fraud detection.
- C. Incorrect.
Streaming serving focuses on continuous data ingestion and processing, but it is often leveraged for processing data streams rather than providing immediate predictions.
- D. Incorrect.
Offline serving refers to scenarios where predictions are pre-computed and stored for later use, which is not suited for real-time decision-making.