Home / Data Analyst practice test / Managing Data

Free · 5 questions with explanations

Managing Data: Databricks Data Analyst Associate Practice Questions

Exam-style questions on Managing Data. Pick your answer, then open the explanation to see why it's right — and why the other options are wrong.

Updated November 10, 2026

1 Managing Data

A data governance team is tagging tables in Unity Catalog to support a new data classification initiative. A data steward needs to add a tag with key pii_category and value contact_info to the customers.profiles.email_address column of an existing Delta table. The steward has been granted APPLY TAG on the table, USE SCHEMA on customers.profiles, and USE CATALOG on customers, but is NOT the table owner. Additionally, a junior engineer asks whether they can tag multiple columns on the same table in a single SQL ALTER TABLE statement. Which of the following statements is CORRECT?

  1. AThe steward is not able to add the tag: APPLY TAG on a table covers the table-level object, and a separate APPLY TAG ON COLUMN privilege has to be granted for each column to be tagged. The junior engineer, however, can tag several columns in one ALTER TABLE statement with a comma-separated column list, which is the recommended approach for bulk column tagging as it reduces transaction overhead on the Unity Catalog metastore.
  2. BThe steward's privileges are sufficient: APPLY TAG on the table covers table-level and column-level tags, and together with USE SCHEMA and USE CATALOG it forms the documented minimum for a non-owner. The tag 'pii_category' = 'contact_info' is added to email_address in Catalog Explorer or with an ALTER TABLE statement. The junior engineer has to tag each column in a separate ALTER TABLE statement, unlike COMMENT, which accepts several columns.
  3. CThe steward's privileges are not sufficient: USE SCHEMA and USE CATALOG are navigation permissions that workspace users receive by default on accessible catalogs, and they play no part in the authorization check for tagging. Column tags are applied by the table owner, so a non-owner is given table ownership temporarily, applies the tags, and has ownership transferred back. The junior engineer can then tag several columns in one statement.
  4. DThe steward's privileges are sufficient, and tag keys are case-insensitive and stored in lowercase, so pii_category and PII_CATEGORY are treated as one tag. A table has no cap on its number of column tags, and the junior engineer can apply up to 50 tags across multiple columns in a single ALTER TABLE statement by adding a WHERE clause that filters the target columns by a name pattern, which avoids one statement per column.
  5. EThe steward can add the tag, and Unity Catalog encrypts the value contact_info with AES-256 before storage, as tag values on PII-related columns are classed as sensitive governance metadata; the values are then visible to users holding the VIEW ENCRYPTED METADATA privilege, while tag keys stay in plain text. The junior engineer's bulk statement succeeds once the steward approves the batch in the tag approval workflow.
Show answer & explanation

Correct answer: B

WHY B: The documentation states that a non-owner adding tags to a Unity Catalog securable needs three privileges: APPLY TAG on the object, USE SCHEMA on the parent schema, and USE CATALOG on the parent catalog. The steward has all three, and APPLY TAG on a table covers both table-level and column-level tagging for that table. Tagging can be done in Catalog Explorer (on the column) or with SQL ALTER TABLE. On bulk tagging the documentation says: 'You cannot assign tags to multiple columns in a single ALTER TABLE command. You must assign tags to each column separately. This differs from the COMMENT clause, which does support multiple columns in one command.' WHY NOT A: APPLY TAG on a table does cover column-level tagging for that table; there is no separate APPLY TAG ON COLUMN privilege. A comma-separated multi-column tag statement is also not possible: the documentation rules out tagging multiple columns in one ALTER TABLE command. WHY NOT C: USE SCHEMA and USE CATALOG are documented as required parts of the three-privilege minimum for a non-owner, not as mere navigation rights, and they are not granted to users by default: Unity Catalog works on a least-privilege model. Ownership is not needed, and multi-column tagging in one statement is not supported. WHY NOT D: Tag keys are case sensitive, so pii_category and PII_CATEGORY are two distinct tags. There are caps: a table can have at most 1,000 column tags across its columns, and a single securable object supports at most 50 tags. ALTER TABLE tagging has no WHERE clause for filtering columns by name pattern. WHY NOT E: Unity Catalog does not encrypt tag values, whether or not they sit on PII-related columns. The documentation warns: 'Tag data is stored as plain text and may be replicated globally. Do not use tag names, values, or descriptors that could compromise the security of your resources.' There is no VIEW ENCRYPTED METADATA privilege, no AES-256 tag encryption, and no tag approval workflow for bulk tagging.

2 Managing Data

A data analyst at a retail company wants to confidently use only high-quality, authoritative datasets for their quarterly revenue report. Their Databricks workspace is enabled for Unity Catalog. Which combination of Unity Catalog features and behaviors BEST supports the analyst in discovering trustworthy datasets, confirming their quality status, and querying them alongside internal tables?

  1. AThe analyst opens the Delta transaction log of each candidate table and checks the _delta_log directory for a JSON quality certificate entry appended by a data steward; tables whose log contains the entry are certified. Queries on certified tables then add a CERTIFIED ONLY keyword, which the Databricks Runtime uses to keep certified and uncertified data from being mixed in one SELECT statement.
  2. BThe analyst uses the workspace search bar or Catalog Explorer to browse tables across catalogs, filters by the 'Certified' endorsement that data stewards apply to tables meeting quality standards, and reads each table's description for context. The chosen tables are then queried with the standard three-part name (catalog.schema.table) in the SQL Editor or a notebook and joined with internal tables; certification is a trust signal, not a query restriction.
  3. CCertified datasets are stored in a dedicated system catalog named certified_data that Databricks provisions when a workspace is created. To query one, the analyst runs USE CATALOG certified_data before the SELECT; a statement that references a table from another catalog alongside a certified_data table raises a cross-catalog certification error, so such comparisons are written as separate queries and joined in Python.
  4. DUnity Catalog does not offer certification or quality endorsements at the table level; certification is a Databricks Marketplace feature for externally published datasets. Internal tables can carry a user-defined tag such as 'certified=true', but that tag is handled like other metadata tags and gets no special treatment in Catalog Explorer or search, so the analyst confirms quality by asking each table owner directly.
  5. EThe analyst asks a workspace administrator to run the CLI command databricks unity-catalog certify --table catalog.schema.table --force-quality-scan on each candidate table. The command starts a data profiling job that scores each column on a 0–100 completeness scale; tables scoring above 80 are promoted to the system.certified schema for querying, and the others are flagged with an UNCERTIFIED comment in INFORMATION_SCHEMA.
Show answer & explanation

Correct answer: B

WHY B: Unity Catalog supports discovery through Catalog Explorer and the workspace search bar, where data stewards apply 'Certified' endorsements (also called quality labels or certifications) to tables validated as authoritative and high-quality. The endorsement appears as a visual indicator in Catalog Explorer next to the table's description, owner, tags and documentation, so analysts identify trustworthy datasets without opening data files or reading logs. The table is then queried with the standard three-part namespace (catalog.schema.table), the same syntax as any registered table. Certification is a discoverability and governance feature, not a query restriction: certified tables are queried like uncertified ones and can be joined freely with other Unity Catalog tables in one SQL statement. WHY NOT A: Parsing the _delta_log directory is not a supported or documented way to identify certified datasets. Delta transaction logs record DML/DDL operations, schema evolution and file statistics and have no concept of a quality certificate entry. The CERTIFIED ONLY keyword does not exist in Databricks SQL. WHY NOT C: There is no system-provisioned certified_data catalog; certified tables stay in their normal catalog and schema, and the endorsement can be applied to a registered table in any catalog. No cross-catalog certification restriction exists: Unity Catalog allows joining tables across catalogs in a single SQL query with the three-part namespace. WHY NOT D: Unity Catalog does support quality endorsements and certification labels at the table level, a documented feature available through Catalog Explorer and the REST API. User-defined tags can also be applied, but the 'Certified' endorsement is a distinct, first-class UI feature with dedicated treatment in search and Catalog Explorer, not an arbitrary custom tag. WHY NOT E: No databricks unity-catalog certify CLI command exists. Certification is managed through the Catalog Explorer UI or the REST API via the table properties endpoint, not through a CLI quality-scan pipeline, and tables are not moved to a system.certified schema on the basis of profiling scores.

3 Managing Data

A data engineer wrote a notebook six months ago that reads from gold.finance.revenue_summary and writes to gold.reporting.quarterly_kpis. A colleague now opens the quarterly_kpis table in Catalog Explorer to trace where the data originally came from. They also want to understand whether column-level lineage is captured. Which combination of statements CORRECTLY describes Unity Catalog's lineage behavior for this scenario?

  1. AUnity Catalog captures lineage for tables that a workspace administrator has registered for lineage tracking with ALTER TABLE ... ENABLE LINEAGE, and tables created before that step have no retroactive history. If lineage was not enabled six months ago, the colleague will see no upstream table for quarterly_kpis. Column-level lineage is available for tables stored in the Iceberg format, while Delta tables get table-level lineage.
  2. BUnity Catalog retains lineage with no time-based expiration: the link between source and target stays in the metastore until an administrator runs ALTER TABLE quarterly_kpis DROP LINEAGE, so the colleague will see revenue_summary upstream. Column-level lineage is captured for each transformation type, including user-defined functions, path-based references such as select * from delta.'s3://...', and RDD-based operations.
  3. CUnity Catalog captures lineage for queries run from the Databricks SQL editor and AI/BI dashboards, while notebook queries using Python or Scala DataFrames are excluded: notebooks run in the classic compute plane, which is isolated from the lineage capture service in the serverless compute plane. The colleague will see quarterly_kpis as a lineage endpoint with no upstream connection from the notebook, at table level or at column level.
  4. DUnity Catalog aggregates lineage within a single workspace, so if the notebook ran in a different workspace from the one the colleague is using, the upstream table is not shown even when both workspaces share one metastore. Column-level lineage is captured, but a data steward first has to annotate the column-to-column mappings in the Catalog Explorer UI before the graph is populated; it is not inferred from the queries themselves.
  5. EUnity Catalog captures runtime lineage for queries on tables registered in the metastore with no configuration, across Python, SQL, Scala and R through Spark DataFrame or Databricks SQL interfaces. The colleague opens quarterly_kpis in Catalog Explorer, selects the Lineage tab and clicks 'See Lineage Graph' to find revenue_summary upstream. Column-level lineage is captured too, and lineage is retained for one year, so a run from six months ago still appears.
Show answer & explanation

Correct answer: E

WHY E: Unity Catalog lineage needs no opt-in or configuration: it is captured at runtime whenever queries run on Databricks against Unity Catalog-registered tables through Spark DataFrame or Databricks SQL interfaces. The documentation says lineage is 'supported for all languages' and captured 'down to the column level', and clicking a column in the graph shows the upstream columns it was derived from. Opening the table in Catalog Explorer, choosing the Lineage tab and 'See Lineage Graph' is the documented procedure. Lineage data is retained for a rolling one-year window, so the link between revenue_summary and quarterly_kpis from a run six months ago is still visible. WHY NOT A: Unity Catalog does not require lineage enrollment with ALTER TABLE ... ENABLE LINEAGE; capture happens for all qualified workloads without setup. Column-level lineage is supported for Delta tables and has no format-based restriction to Iceberg. WHY NOT C: The documentation states lineage is 'supported for all languages' and captured via 'Spark DataFrame (for example, Spark SQL functions that return a DataFrame) or Databricks SQL interfaces such as notebooks or the SQL query editor'. Notebooks using DataFrames or Spark SQL are covered; the classic versus serverless compute distinction does not exclude notebook lineage. WHY NOT D: Lineage is 'aggregated across all workspaces attached to a Unity Catalog metastore', so a run from another workspace on the same metastore shows up in the colleague's view. Column-level lineage is inferred from query plans and needs no manual annotation by a data steward. WHY NOT B: Lineage data is retained for one year (a rolling window), not indefinitely, and there is no ALTER TABLE ... DROP LINEAGE command. The documentation also lists cases where column-level lineage is not captured: UDFs (which can obscure column mapping), path-based table references such as select * from delta.'s3://...', RDD-based operations, global temp views, and checkpointed datasets.

4 Managing Data

A data analyst needs to find all tables in their organization's Databricks Unity Catalog environment that are relevant to customer purchase behavior, then query one of those tables alongside a separate internal order-fulfillment table that lives in a different catalog. The analyst's account has SELECT on the target tables and USE CATALOG and USE SCHEMA on both parent objects. Which approach CORRECTLY demonstrates Unity Catalog's discovery and cross-catalog querying capabilities?

  1. AUnity Catalog does not run a cross-catalog join in a single SELECT statement. The analyst finds candidate tables in Catalog Explorer, uses CREATE TABLE AS SELECT to copy one table into the catalog of the other, runs the JOIN on the two co-located tables, and drops the copy afterwards. This catalog isolation is enforced at the SQL warehouse level, and granting additional privileges on either catalog does not lift it.
  2. BThe workspace search bar covers workspace objects such as notebooks, jobs and dashboards, so to search Unity Catalog tables by business context the analyst installs the separately licensed Data Discovery Add-on, which crawls metastore metadata nightly. The cross-catalog JOIN with three-part names then runs provided both catalogs use the same cloud storage region as the SQL warehouse that executes the query.
  3. CWorkspace search indexes table names and schema names but not column names, comments or tags, so the analyst queries INFORMATION_SCHEMA.TABLES in each catalog with a LIKE '%purchase%' filter and compiles the results into one inventory. Before the cross-catalog JOIN, the analyst runs SET SEARCH_PATH TO catalog1, catalog2, which tells the SQL warehouse which catalogs to make available in the current session.
  4. DThe analyst first registers a federation link with CREATE CATALOG LINK retail_to_logistics BETWEEN retail AND logistics; without the link, the SQL parser rejects a query that names tables from two catalogs and returns a 'CatalogLinkNotFound' error. Once the link exists, tables are found through search and joined with a two-part prefix syntax such as [retail]customers.purchase_history in place of the three-part name.
  5. EThe analyst types keywords such as 'customer orders' in the workspace search bar, which returns matching tables, views and schemas across the catalogs visible to the account, with matches in table names, column names, comments and tags. The join then uses three-part names: SELECT c.customer_id, o.fulfillment_status FROM retail.customers.purchase_history c JOIN logistics.orders.fulfillment o ON c.order_id = o.order_id, with no data copy or extra privilege.
Show answer & explanation

Correct answer: E

WHY E: Workspace search is a first-class discovery tool that indexes table names, schema names, column names, table and column comments, and user-defined tags, so keyword search works across the catalogs the user can access. Unity Catalog's design supports cross-catalog SQL with the three-part namespace (catalog.schema.table): no USE CATALOG restriction prevents referencing another catalog in the same SELECT, and with USE CATALOG, USE SCHEMA and SELECT on both tables the JOIN executes normally. This is fundamental to the lakehouse pattern in which domain data sits in several catalogs (for example retail, logistics, finance) but is analyzed together. WHY NOT A: Cross-catalog queries in a single SELECT are fully supported, and no SQL warehouse-level isolation policy blocks multi-catalog JOINs. Copying data between catalogs before joining would defeat the purpose of the three-part namespace. WHY NOT C: Workspace search indexes more than table and schema names: column names, descriptions, comments and tags are searchable too. SET SEARCH_PATH TO is not Databricks SQL syntax for cross-catalog queries; the fully qualified three-part name is sufficient and no session configuration is required. WHY NOT D: No CREATE CATALOG LINK command exists in Databricks SQL or Unity Catalog. Querying external databases such as MySQL or PostgreSQL uses Lakehouse Federation connections (CREATE CONNECTION), but two Unity Catalog tables in different catalogs share one metastore and need no link; the three-part name is sufficient. WHY NOT B: The workspace search bar includes Unity Catalog data assets as a built-in feature with no add-on license. Both catalogs do belong to the same Unity Catalog metastore (the standard setup within an account in a region), but there is no cloud storage region co-location requirement for cross-catalog JOINs on a SQL warehouse.

5 Managing Data

A data analyst is cleaning a Unity Catalog Delta table named silver.customers.profiles. The table has a signup_date column (DATE type) that contains some NULL values, and an email column (STRING type) where some rows contain an empty string '' instead of a valid email address. The analyst wants to: (1) replace NULL signup_date values with the date '2020-01-01' in a query result (without modifying the underlying table), and (2) permanently delete all rows where email is either NULL or an empty string from the underlying table. Which SQL approach CORRECTLY accomplishes both requirements?

  1. ARequirement 1: REPLACE(signup_date, NULL, '2020-01-01') in the SELECT clause; REPLACE is the standard SQL function for substituting a default for NULL values in a query result. Requirement 2: UPDATE silver.customers.profiles SET email = NULL WHERE email = '' to represent both forms of missing email as NULL, followed by DELETE FROM silver.customers.profiles WHERE email IS NULL to remove those rows from the underlying Delta table.
  2. BRequirement 1: COALESCE(signup_date, DATE '2020-01-01'), or equivalently NVL(signup_date, '2020-01-01'), in the SELECT clause; COALESCE returns its first non-NULL argument, so existing signup_date values are kept in the result and the underlying table is left unmodified. Requirement 2: DELETE FROM silver.customers.profiles WHERE email IS NULL OR email = '', a single statement that removes the rows with a NULL or empty email from the Delta table.
  3. CRequirement 1: IFNULL(signup_date, CURRENT_DATE()) in the SELECT clause; a static value such as '2020-01-01' is not accepted, as DATE literals are not valid as the second argument of IFNULL in Databricks SQL. Requirement 2: TRUNCATE TABLE silver.customers.profiles, then re-insert the rows from a subquery that filters out NULL and empty emails; this is the recommended route, as DELETE is not supported on Unity Catalog Delta tables.
  4. DRequirement 1: COALESCE(signup_date, '2020-01-01') after running ALTER TABLE silver.customers.profiles ALTER COLUMN signup_date SET DEFAULT '2020-01-01'; COALESCE reads the column DEFAULT and returns NULL for NULL input when no DEFAULT is set. Requirement 2: DELETE FROM silver.customers.profiles WHERE email IS NULL OR email = '', run after disabling ACID transactions with SET TBLPROPERTIES ('delta.enableAcid' = 'false') for the delete to commit.
  5. ERequirement 1: CASE WHEN signup_date = NULL THEN DATE '2020-01-01' ELSE signup_date END in the SELECT clause; this is preferred over COALESCE and NVL, which write the substituted value back to the underlying column unless the query is wrapped in a BEGIN TRANSACTION block that is rolled back. Requirement 2: DELETE FROM silver.customers.profiles WHERE email IS NULL OR email = '', a single statement on the Delta table.
Show answer & explanation

Correct answer: B

WHY B: Both requirements are met with standard Databricks SQL. (1) COALESCE(signup_date, DATE '2020-01-01') returns the first non-NULL value in its argument list: the actual date when signup_date is not NULL, the default '2020-01-01' when it is. COALESCE is a projection function used in the SELECT output and does not modify the underlying table; NVL is functionally equivalent and also valid in Databricks SQL. (2) DELETE FROM silver.customers.profiles WHERE email IS NULL OR email = '' is a standard Delta Lake DML statement that permanently removes the rows matching the predicate. Delta tables in Unity Catalog fully support DELETE, UPDATE, and MERGE as ACID DML operations with no special configuration. WHY NOT A: REPLACE(string, search, replace) is a string function that replaces occurrences of a substring; it does not operate on NULL values, is not equivalent to COALESCE, and on a DATE column would cause a type mismatch error. The two-step UPDATE then DELETE is a valid alternative for requirement 2, though less efficient than a single DELETE with an OR condition. WHY NOT C: IFNULL(signup_date, CURRENT_DATE()) is valid Databricks SQL but substitutes today's date, not the fixed date '2020-01-01' the question requires, and DATE literals can be used as the second argument of IFNULL and COALESCE. TRUNCATE TABLE removes all rows unconditionally and is not a selective deletion tool; Delta tables in Unity Catalog fully support DELETE. WHY NOT D: COALESCE evaluates its arguments at runtime in the query and is independent of column DEFAULT definitions, so no DEFAULT has to be set first. ACID transactions cannot be disabled with delta.enableAcid = 'false': ACID support is a fundamental, non-optional property of Delta tables and no such table property exists. WHY NOT E: A comparison with = NULL evaluates to NULL, never true, so this CASE expression leaves the NULL dates unreplaced; the test has to be IS NULL. COALESCE and NVL are pure expression functions that evaluate at query runtime and do not modify underlying column data, so they can be used in SELECT clauses, WHERE clauses and expressions without a transaction; COALESCE is the idiomatic, concise choice.

Take the full Data Analyst practice test →