MLS-C01 Question 21
Single answerA company is building a machine learning pipeline to process user activity data. The pipeline must handle a continuous stream of data to detect fraudulent transactions in real-time. Which job style and job type should the company use for this use case?
- A
Batch processing with a scheduled job
- B
Real-time processing with a streaming job
- C
Batch processing with an on-demand job
- D
Real-time processing with a batch load job
Show answer and explanation
Correct answer: B
Explanation
For detecting fraudulent transactions in real-time, the pipeline must process data as it is generated. Streaming jobs are designed for real-time or near real-time data processing, making 'Real-time processing with a streaming job' the correct choice. Batch processing, whether scheduled or on-demand, introduces latency that is unsuitable for real-time applications.
- A. Incorrect.
Batch processing with a scheduled job is not suitable for real-time use cases because it processes data in bulk at predefined intervals, which introduces significant latency and is unsuitable for detecting fraud in real-time.
- B. Correct.
Real-time processing with a streaming job is the correct choice because it allows continuous ingestion and processing of data streams as they arrive, enabling real-time fraud detection.
- C. Incorrect.
Batch processing with an on-demand job is also unsuitable for real-time use cases, as it typically processes a large dataset only when explicitly triggered, leading to delays.
- D. Incorrect.
Real-time processing with a batch load job is incorrect because batch load jobs are not designed for continuous or real-time data ingestion.