Associate Data Practitioner

By Google · Question Mode

QuestionQ52

Data Analysis and Presentation

You built a customer purchase-propensity model with BigQuery ML six months ago. You want to compare the current serving data against the historical serving data to decide whether the model needs retraining. What should you do?

Explanation

Data drift identifies significant changes in feature distributions between two serving-data sets over time. BigQuery ML’s ML.VALIDATE_DATA_DRIFT is specifically designed to compare current serving data with historical serving data and detect anomalous differences that can indicate a need for retraining.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!