Databricks Data Analyst Sample Questions

About these questions

These five questions are original. We wrote them to match the style of the exam outline, with scenarios and plausible wrong answers, but they are not real exam questions, and the real exam will word things differently. They touch five of the nine sections: queries, query analysis, Genie spaces, dashboards, and securing data. Our practice tests cover all nine.

How to use these

Try each question without looking at your notes, and pick an answer before you open the explanation. Then read the explanation even when you were right, because the reason the closest wrong option fails is where the learning is. Note any section where you slipped and revisit it on the domains explained page.

Question 1 (Executing Queries Using Databricks SQL)

A support team report lists customers with `ORDER BY last_contact DESC`, so the most recent contact comes first. Some customers have never been contacted, so `last_contact` is `NULL` for them. The query has no `NULLS FIRST` or `NULLS LAST` clause. Where do the never-contacted customers appear?

  • A. At the top of the result, before the most recent dates.
  • B. They are removed from the result.
  • C. At the bottom of the result, after the oldest dates.
  • D. Scattered through the result in the order the rows were stored.
Show the answer and explanation

Correct answer: C. In Databricks SQL, `NULL` values sort first in ascending order and last in descending order when you do not say otherwise. This report sorts descending, so the empty values end up at the bottom. Option A is the closest wrong answer: it would be right for `ASC`, which is why switching a report from oldest-first to newest-first moves the empty values. `ORDER BY` never removes rows, and it never leaves rows in storage order. When the column can be `NULL`, write `NULLS FIRST` or `NULLS LAST` so the result does not depend on a default.

Question 2 (Analyzing Queries)

An analyst overwrote a table with the wrong file a few minutes ago and wants to restore the previous contents with Delta time travel. She runs `DESCRIBE HISTORY` on the table to find the right version number. Where does the overwrite appear in the output?

  • A. In the first row, because the history lists the newest operation first.
  • B. In the last row, because the history starts at version 0.
  • C. In the row for the table creation, because overwrites replace that entry.
  • D. In no row, because history only lists schema changes.
Show the answer and explanation

Correct answer: A. `DESCRIBE HISTORY` returns one row per commit in reverse chronological order, so the latest operation is on top. Each row has a version number, timestamp, user, operation, and metrics. The analyst reads the overwrite’s version, then queries the version just before it, for example with `VERSION AS OF`. Option B is the closest wrong answer. Version numbers do start at 0, but the listing shows the highest number first. Overwrites add a new entry rather than replacing one, and history records data writes, not only schema changes.

Question 3 (Developing, Sharing, and Maintaining AI/BI Genie Spaces)

A finance manager has `CAN RUN` permission on a Genie space, and the space’s SQL warehouse is running. Every question the manager asks returns no data, while the space owner gets answers. What is the most likely cause?

  • A. The space has no sample questions, so Genie cannot answer anyone.
  • B. The `CAN RUN` permission does not let a user ask questions, so it must be raised to `CAN MANAGE`.
  • C. The manager does not have `SELECT` on the Unity Catalog tables behind the space.
  • D. Genie runs every query as the space owner, and the owner’s session has expired.
Show the answer and explanation

Correct answer: C. Using a Genie space needs two things: a permission on the space itself, and access to the underlying data in Unity Catalog. Genie queries run as the person asking, so the manager also needs `SELECT` on the tables, plus `USE CATALOG` and `USE SCHEMA` on their parents. Option B is the closest wrong answer, but `CAN RUN` already allows asking questions. Option D is wrong because queries run under each user’s own identity, not the owner’s. Missing sample questions would make answers worse, not empty for one user only.

Question 4 (Working with Dashboards and Visualizations in Databricks)

An operations lead wants a chart showing how delivery times, measured in minutes, are spread across their range, to see whether most deliveries are fast or slow. Which visualization type fits best?

  • A. Bar chart with one bar per delivery
  • B. Scatter chart of delivery time against order ID
  • C. Pie chart with one slice per minute value
  • D. Histogram of delivery time
Show the answer and explanation

Correct answer: D. A histogram groups one numeric measure into ranges and shows how many values fall into each, which is exactly a distribution question. Option A is the closest wrong answer. Bar charts compare a measure across categories, and one bar per delivery would be unreadable and would not show the shape of the data. A scatter chart looks for a relationship between two measures, and order ID is only an identifier. A pie chart with one slice per minute would be cluttered and says nothing about spread.

Question 5 (Securing Data)

A nightly job refreshes a reporting table from governed Unity Catalog data. It currently runs under the personal account of an engineer who has just left the company. Which identity should run the job from now on?

  • A. A shared login that the whole team knows the password to.
  • B. A service principal that has been granted only the privileges the job needs.
  • C. A group, because groups can sign in on behalf of the job.
  • D. The personal account of the engineer who inherits the job.
Show the answer and explanation

Correct answer: B. Service principals are the identity type meant for jobs, scripts, and other automated tools. They are not tied to a person’s employment, and you can grant them just the privileges the job needs. Option D is the closest wrong answer: it works today but recreates the same problem when that person moves on, and it ties the job’s access to their wider rights. A shared login destroys any audit trail. Groups are used to grant access to many identities at once, but they do not authenticate.

More guides for DATABRICKS-DATA-ANALYST-ASSOCIATE

Next step

If a question surprised you, the domains explained page shows where it fits. For more of this style across all nine sections, with per-option feedback, see our Databricks Data Analyst Associate practice tests.

Try these practice exams