Databricks Data Analyst

Before any analysis you have to trust the data. You search and query certified datasets, apply governed tags to assets, and clean tables with SQL: types, nulls, duplicates and inconsistent values.

• Searching and previewing certified datasets
• Governed tags and how to apply them
• Handling nulls, duplicates and bad values
• Casting types and standardizing text
• Documenting what you changed and why

You build A cleaned table with tags and a change note that a colleague could follow.

Analysts rarely start with the data they need. You explain how data gets into the platform, from file uploads and Auto Loader to Delta Sharing, APIs and managed connectors, and choose the route that fits a source.

• Uploading files for one-off analysis
• Auto Loader for files that keep arriving
• Delta Sharing for data from other organizations
• Managed connectors and API sources
• Choosing a route for a stated source

You build An uploaded and a shared dataset combined in one query, with a note on how each arrived.

SQL is the analyst’s main tool. You write joins, set operations and aggregates, query federated sources in place, and learn how a warehouse is sized and started so that your query runs on the right compute.

• Joins and their effect on row counts
• Set operations and subqueries
• Aggregates, grouping and window functions
• Querying federated sources
• Warehouse types, sizes and startup

You build A set of queries that answers five business questions across joined tables.

Repeated queries should not repeat work. You create materialized views for expensive aggregations, manage tables in SQL, travel back in time to compare versions of data, and use Genie Code to draft and check queries.

• Materialized views and refreshes
• Creating and managing tables in SQL
• Time travel: querying earlier versions
• Restoring a table after a mistake
• Drafting and reviewing SQL with Genie Code

You build A materialized view for a costly report, plus a time travel comparison of two table versions.

A slow query costs money and patience. You read query history and the query profile to see where time goes, apply fixes such as better filters and caching, and know when Photon and predictive optimization already help.

• Query history and insights
• Reading a query profile
• Filters, joins and what makes them slow
• Result and disk caching
• Photon and predictive optimization

You build A before-and-after profile of a slow query, with the change that sped it up.

How tables are shaped decides how easy questions are to ask. You compare star, snowflake and data vault schemas, relate them to the medallion layers, and choose a model that makes the analysis you need straightforward.

• Facts, dimensions and grain
• Star and snowflake schemas compared
• Data vault: hubs, links and satellites
• Medallion layers and where analysis happens
• Choosing a model for a reporting need

You build A star schema design for a sales question, with the queries it makes simple.

A dashboard is a set of decisions about what to show. You build AI/BI dashboards on several datasets, add parameters and filters, and choose chart types that answer the question instead of decorating it.

• Datasets and how a dashboard uses them
• Choosing a chart for a question
• Parameters, filters and cross-filtering
• Combining multiple datasets on one page
• Layout and readability

You build A dashboard on three datasets with parameters and charts chosen for stated questions.

A dashboard is only useful if it is current and reaches the right people. You schedule refreshes, set alerts on thresholds, and share with the right permissions so viewers see what they should, and nothing more.

• Refresh schedules and their cost
• Alerts on thresholds and conditions
• Sharing with users and groups
• Embedded credentials and data access
• Subscriptions and notifications

You build A scheduled dashboard with an alert and a sharing setup for two audiences.

Genie lets colleagues ask questions in plain language, but only if it is set up well. You create a space with sample questions and trusted assets, share it, and improve it using user feedback and benchmark questions.

• Choosing tables and writing instructions
• Sample questions and trusted assets
• Sharing a space with the right people
• Reading feedback and failed questions
• Benchmarks to measure improvement

You build A Genie space with instructions, trusted queries and a benchmark that tracks its accuracy.

Analysts handle data other people care about. You use Unity Catalog roles and sharing settings, work with the three-level namespace and ownership, and protect personal data in queries and outputs.

• The three-level namespace
• Roles, privileges and ownership
• Sharing settings and their limits
• Recognizing and protecting PII
• Masked views for analysts

You build A secured analytical schema with a masked view and a privilege summary.

You answer a business question from beginning to end: find and clean the data, model it, query it, build a dashboard and a Genie space, secure it, and review its performance.

• Turning a question into a data need
• Finding, cleaning and modeling the data
• Dashboard and Genie space together
• Securing and sharing the product
• Checking performance and cost

You build A dashboard, a Genie space and a secured data model that answer one business question.