PROFESSIONAL-MACHINE-LEARNING-ENGINEER Question #221: Real Exam Question with Answer & Explanation

Question

You have trained a model by using data that was preprocessed in a batch Dataflow pipeline. Your use case requires real-time inference. You want to ensure that the data preprocessing logic is applied consistently between training and serving. What should you do?

Options

• APerform data validation to ensure that the input data to the pipeline is the same format as the
• BRefactor the transformation code in the batch data pipeline so that it can be used outside of the
• CRefactor the transformation code in the batch data pipeline so that it can be used outside of the
• DBatch the real-time requests by using a time window and then use the Dataflow pipeline to

Explanation

To ensure consistent data preprocessing for models trained with batch Dataflow and served for real-time inference, the transformation logic must be reusable across both environments.

Common mistakes.

• A. Data validation ensures input quality but does not guarantee that the same preprocessing transformations are applied consistently between training and serving.
• C. This choice is identical to B, implying the same correct technical reason for refactoring the code.
• D. Batching real-time requests and using Dataflow for processing would introduce significant latency, making it unsuitable for real-time inference requirements.

Concept tested. Training-serving skew prevention, data preprocessing consistency

Reference. https://cloud.google.com/architecture/overview-ml-training-serving-skew

No community discussion yet for this question.