Identify How Streaming Inference Is Performed with Delta Live Tables MACHINE-LEARNING-ASSOCIATE Practice Questions (Page 2)

Questions 6–10

1. 6foundation · easy

   In a DLT streaming inference pipeline, how can you apply a trained ML model to streaming data using Python?

   • ABy using the SQL function ML_PREDICT which is built into DLT
   • BBy using the MLflow model registry directly in the DLT pipeline definition
   • CBy using the spark.ml library's PipelineModel to transform data automatically
   • DBy using spark.udf.register to register a Python function that calls the model's predict method, then applying it in a SELECT statement
   Reveal Solution Discussion Select an answer first
2. 7foundation · easy

   Which DLT feature allows you to define and enforce data quality constraints on streaming inference outputs?

   • AMaterialized views
   • BCheckpoints
   • CStreaming tables
   • DExpectations
   Reveal Solution Discussion Select an answer first
4. 8application · medium

   A team is evaluating whether to use DLT for streaming inference. They have a trained model and a Kafka source. They want to minimize operational overhead. What is the primary benefit of using DLT for this workload?

   • ADLT automatically manages the streaming pipeline, including checkpointing and recovery.
   • BDLT automatically retrains the model on new data.
   • CDLT eliminates the need for a Kafka cluster.
   • DDLT guarantees zero latency for all streaming workloads.
   Reveal Solution Discussion Select an answer first
5. 9foundation · easy

   What is the primary benefit of using Delta Live Tables (DLT) for streaming inference compared to batch inference?

   • ADLT automatically trains and retrains machine learning models on incoming data.
   • BDLT guarantees exactly-once processing for all streaming sources without any configuration.
   • CDLT processes data incrementally as it arrives, enabling lower-latency predictions without waiting for a full batch.
   • DDLT eliminates the need for a separate model registry by storing models directly in tables.
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   In a DLT streaming inference pipeline, what is the purpose of watermarking when performing windowed aggregations?

   • AIt defines the maximum delay allowed for late-arriving data to be included in a window.
   • BIt specifies the size of the window in seconds.
   • CIt ensures exactly-once processing semantics.
   • DIt determines the number of partitions for the streaming query.
   Reveal Solution Discussion Select an answer first