nerdexam
Google

PROFESSIONAL-MACHINE-LEARNING-ENGINEER · Question #57

You have a demand forecasting pipeline in production that uses Dataflow to preprocess raw data prior to model training and prediction. During preprocessing, you employ Z-score normalization on data…

The correct answer is B. Translate the normalization algorithm into SQL for use with BigQuery. Translating the Z-score normalization algorithm into SQL and executing it directly in BigQuery is the most efficient approach. The data already lives in BigQuery, so performing the normalization in-place using SQL eliminates the need to export data to an external system (like…

Submitted by wei.xz· Apr 18, 2026Data processing and feature engineering

Question

You have a demand forecasting pipeline in production that uses Dataflow to preprocess raw data prior to model training and prediction. During preprocessing, you employ Z-score normalization on data stored in BigQuery and write it back to BigQuery. New training data is added every week. You want to make the process more efficient by minimizing computation time and manual intervention. What should you do?

Options

  • ANormalize the data using Google Kubernetes Engine.
  • BTranslate the normalization algorithm into SQL for use with BigQuery.
  • CUse the normalizer_fn argument in TensorFlow's Feature Column API.
  • DNormalize the data with Apache Spark using the Dataproc connector for BigQuery.

How the community answered

(60 responses)
  • A
    10% (6)
  • B
    82% (49)
  • C
    3% (2)
  • D
    5% (3)

Explanation

Translating the Z-score normalization algorithm into SQL and executing it directly in BigQuery is the most efficient approach. The data already lives in BigQuery, so performing the normalization in-place using SQL eliminates the need to export data to an external system (like Dataflow, GKE, or Dataproc), process it, and write it back - each of those steps adds latency, network overhead, and operational complexity. BigQuery is massively parallel and optimized for large-scale analytical transformations, making SQL-based normalization fast and cost-effective. This minimizes both computation time and manual intervention since BigQuery queries can be easily automated and scheduled. Options A, C, and D all involve moving data outside of BigQuery unnecessarily.

Topics

#Data preprocessing#BigQuery SQL#ML pipeline optimization#Feature engineering

Community Discussion

No community discussion yet for this question.

Sign up to join the discussion
Full PROFESSIONAL-MACHINE-LEARNING-ENGINEER Practice