Advanced Feature Store Concepts MACHINE-LEARNING-PROFESSIONAL Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   A logistics company is building a model to predict delivery delays. They have a feature table with 'average traffic speed' updated every 5 minutes. The training data contains delivery records with timestamps. They use the Feature Lookup API with the timestamp_lookup_key set to the delivery timestamp. What is the primary benefit of this approach?

   • AIt reduces the size of the training dataset by removing duplicate records.
   • BIt speeds up the training process by using precomputed features.
   • CIt prevents data leakage by using only traffic data available at the time of delivery.
   • DIt ensures that the feature table is updated in real-time.
   Reveal Solution Discussion Select an answer first
2. 7expert · hard

   A company is designing a streaming pipeline to ingest real-time sensor data and compute features for a predictive maintenance model. They need to ensure that the features are available for both training and real-time inference. They plan to use a streaming pipeline that writes to a Delta table and then create an online table from that Delta table. What is a key design consideration to ensure consistency between training and inference?

   • AUse a batch pipeline instead of streaming to simplify the architecture.
   • BUse a schema that does not include timestamps to reduce complexity.
   • CInclude a timestamp in the streaming data and use it for point-in-time lookups in training.
   • DWrite directly to the online table from the streaming pipeline.
   Reveal Solution Discussion Select an answer first
4. 8application · medium

   A fraud detection team is building a training dataset from historical transactions. Each transaction has an event timestamp. They want to join a feature table that stores rolling 24-hour transaction counts per customer, updated hourly. Which Feature Lookup API parameter ensures that each training row gets the feature value that was current at the transaction's event time, rather than the latest value?

   • Afeature_name
   • Blookup_key
   • Ctimestamp_lookup_key
   • Dstart_time
   Reveal Solution Discussion Select an answer first
5. 9foundation · easy

   How does the FeatureEngineering Client enable automation of feature pipelines?

   • ABy deploying the feature pipeline to a serverless environment automatically.
   • BBy providing a visual interface to design pipelines without writing code.
   • CBy automatically generating feature values from raw data without any code.
   • DBy allowing you to define a feature computation function and schedule it to run on a regular basis, updating the feature table.
   Reveal Solution Discussion Select an answer first
6. 10application · medium

   A financial services firm wants to update a feature table with real-time stock price features. They plan to use a streaming pipeline that ingests from a Kafka topic and writes to the feature table. Which design consideration is most important to ensure that the feature table remains consistent for both training and serving?

   • AWrite directly to the online table without a Delta table.
   • BUse a schema that does not include timestamps to reduce complexity.
   • CUse a batch job that runs every hour to update the feature table.
   • DUse a streaming sink that writes to the feature table with a timestamp column for each feature value.
   Reveal Solution Discussion Select an answer first