Home / DE Associate practice test / Data Transformation and Modeling / Question

Databricks Data Engineer Associate · free practice question

In a Lakeflow Spark Declarative Pipeline, a bronze streaming table ingests append-only clickstream events. A downstream Gold object must present hourly revenue…

Exam-style question on Data Transformation and Modeling from the free DE Associate practice test, with the answer and a full explanation below.

Updated November 10, 2026

Data Transformation and Modeling

In a Lakeflow Spark Declarative Pipeline, a bronze streaming table ingests append-only clickstream events. A downstream Gold object must present hourly revenue totals that always reflect the full, current state of the upstream data, including late-arriving corrections. Which object type is the most appropriate for the Gold aggregation, and why?

  1. AA materialized view: it recomputes its result from the current state of the upstream tables on every pipeline update, so late-arriving corrections are folded back into the current hourly revenue aggregates automatically.
  2. BA streaming table, because a streaming table reprocesses the entire history of its source on every pipeline run and rebuilds all of its aggregates from scratch, which is the only way hourly revenue totals stay correct.
  3. CA standard non-materialized view, because it physically persists the aggregated hourly totals to storage and is therefore by far the fastest possible option to serve those same figures to repeated downstream BI queries.
  4. DA streaming table, because materialized views cannot be declared inside a Lakeflow declarative pipeline and are only ever available as standalone objects created directly through the Databricks SQL editor outside a pipeline.
  5. EA temporary view, because temporary views persist across pipeline runs, are automatically governed by Unity Catalog, and continuously keep the hourly totals refreshed as late corrections arrive from upstream.
Answer & explanation

Correct answer: A

WHY A: A materialized view recomputes from the current state of its sources on each update, so aggregations reflect late-arriving and corrected data — the right fit for a Gold summary table. WHY NOT B: Streaming tables process each input row exactly once via a checkpoint; they are for append-style ingestion, and the claim that they rebuild all history every run is false. WHY NOT C: A standard view stores no data — it is evaluated at query time and does not persist results physically. WHY NOT D: Materialized views are first-class objects inside declarative pipelines. WHY NOT E: A temporary view materializes no data and is pipeline-scoped, so it does not persist across runs.

Take the full DE Associate practice test free →

Would you pass today?

This question is one of 45 in the free, timed Databricks Data Engineer Associate practice test. Take it and get a section-by-section score.

Start the free test →