Feature Store Management MACHINE-LEARNING-ASSOCIATE Practice Questions (Page 4)

Questions 16–18

1. 16application · medium

   A company has a real-time fraud detection model that needs to score transactions with sub-second latency. The model uses features from a feature store. Which feature table configuration should they use for online inference?

   • AA cached DataFrame in the model serving environment.
   • BAn offline feature table stored as a Delta table in cloud storage.
   • CAn online feature table backed by a low-latency database such as DynamoDB.
   • DA view that queries the offline table on demand.
   Reveal Solution Discussion Select an answer first
2. 17expert · medium

   A company is migrating from workspace-level feature tables to account-level tables in Unity Catalog. They have multiple workspaces for different business units. What is a key benefit they will gain from this migration?

   • ACentralized governance and the ability to share features across workspaces.
   • BImproved query performance due to indexing.
   • CReduced storage costs because account-level tables use compression.
   • DAutomatic replication of data to all regions.
   Reveal Solution Discussion Select an answer first
3. 18foundation · easy

   When writing data to a feature store table, what does the Feature Store API do to ensure data consistency?

   • AIt deletes all existing data and replaces it with the new DataFrame.
   • BIt only inserts new rows and ignores any rows that already exist.
   • CIt performs upserts based on the primary key, updating existing rows and inserting new ones.
   • DIt requires the user to manually specify whether to update or insert each row.
   Reveal Solution Discussion Select an answer first