Certified Machine Learning ProfessionalDemo

Question

A Machine Learning Engineer previously built a feature table for model training and inference using a batch-mode approach:

Question Image

They have since been told that these features must be available in “real-time,” with latency of approximately a minute. Their manager has told them that a Kafka stream is now available for streaming live data, and they must ingest it and make it available for low-latency feature lookups.

Which modification to their existing code will accomplish this?

A Change the incoming_df to be a dataframe based on a readStream() from the kafka source, the write_table() method will provide a low-latency lookup on this data.
B Change the incoming_df to be a dataframe based on a readStream() from the Kafka source and publish the table as an online table with the streaming option set to True.
C Run a triggered workflow to ingest the Kafka data to a dataframe that they can use with their existing write_table() command.
D Create a custom pyfunc MLflow model which processes results of the Kafka stream for on demand feature calculation.

Explanation

A Kafka readStream() DataFrame enables continuous updates to the offline feature table, while publishing that table to an online store with streaming enabled continuously synchronizes feature updates for low-latency real-time lookups. write_table() alone maintains the feature table but does not create the online serving store required for low-latency access.

Learn more

Certified Machine Learning ProfessionalDemo

AstroTutor

Notes

Readiness

QuestionQ142

Community Discussion