Microsoft Fabric Data and Analytics
OneLake, lakehouses, pipelines, warehouses, Power BI, Real-Time Intelligence and AI in one analytics platform
About this course
A six-module course on building an end-to-end analytics solution in Microsoft Fabric. You start with how Fabric is organised around OneLake, workspaces and capacity, then engineer data in lakehouses with Spark, ingest and transform it with pipelines and dataflows, and serve it through warehouses, semantic models and Power BI.
Later modules add Real-Time Intelligence for streaming data, governance and deployment practices, and the AI layer: data science with MLflow, Copilot across the platform and data agents that answer questions from your data. A capstone on a fictional clinic network ties it together.
What you’ll be able to do
- Explain Fabric’s architecture, OneLake and capacity model, and organise workspaces and domains
- Engineer lakehouse data with Spark notebooks and maintain Delta tables
- Ingest and transform data with pipelines, Dataflow Gen2 and mirroring
- Serve data through warehouses, Direct Lake semantic models and secure Power BI reports
- Build real-time analytics with eventstreams, eventhouses, KQL and alerts
- Govern and deploy Fabric items, and apply data science, Copilot and data agents responsibly
Before you start
What this course assumes you can already do.
- Working SQL and some Python.
- Prior Power BI experience helps but is not required.
The curriculum
Open any module to see its objective, every teaching topic, the practice brief and the checkpoint.
Recaps and reference material are included in the LMS.
PART 1Foundations and Engineering
01Fabric Fundamentals, OneLake and Workspaces
Explain what Fabric unifies, how OneLake stores and shares data, and how workspaces, domains and capacity are organised.
- What Fabric unifies: experiences, items and a single platformConcept
- OneLake: one logical lake, open Delta format and shortcutsConcept
- Workspaces, roles and domainsHands-on
- Capacities, sizes and the capacity metrics appConcept
- Creating shortcuts to external storageHands-on
A workspace plan for a clinic network
Design the workspace and domain structure for a clinic network with operations, finance and data science teams, stating roles for each workspace. Create one workspace, add a lakehouse and a shortcut to an external storage folder, and query it. Check the capacity metrics app after a heavy query and describe what you see.
Knowledge checkpoint with explained answers
02Lakehouse, Spark and Data Engineering
Engineer data in Fabric lakehouses with Spark notebooks, apply a medallion design and keep Delta tables healthy.
- Creating a lakehouse: files, tables and the SQL analytics endpointHands-on
- Notebooks and Spark in Fabric: sessions, environments and librariesHands-on
- Transforming data into Delta tables with PySparkHands-on
- Medallion design in Fabric lakehousesHands-on
- Table maintenance: V-Order, optimisation and vacuumHands-on
- Spark job definitions and scheduled notebooksHands-on
A medallion lakehouse for appointments
Build bronze, silver and gold layers for a synthetic appointments feed containing duplicates, invalid times and late records. Use PySpark notebooks, quarantine failing rows in silver and build a gold table of daily appointments and no-shows by clinic. Schedule the notebooks, run table maintenance and show the gold table queried through the SQL endpoint.
Knowledge checkpoint with explained answers
03Data Factory, Pipelines and Dataflows
Ingest and transform data in Fabric with pipelines, Dataflow Gen2 and mirroring, with control flow, scheduling and alerting.
- Data Factory in Fabric: pipelines versus Dataflow Gen2Concept
- Copy activity and copy jobs for ingestionHands-on
- Transforming with Dataflow Gen2 and Power QueryHands-on
- Pipeline control flow: parameters, ForEach, conditions and notebooksHands-on
- Mirroring operational databases into OneLakeHands-on
- Scheduling, monitoring and failure alertsHands-on
A metadata-driven ingestion pipeline
Build a pipeline that reads a control table of source tables, copies each incrementally into bronze with ForEach, then calls a notebook to build silver. Add a Dataflow Gen2 that cleans a staff roster spreadsheet into the warehouse. Schedule the pipeline, add a failure alert and demonstrate a clean rerun after a deliberate failure.
Knowledge checkpoint with explained answers
PART 2Serving and Real Time
04Warehousing, Power BI and Semantic Models
Serve data through Fabric warehouses and Direct Lake semantic models, and build secure Power BI reports.
- The Fabric warehouse: T-SQL, cross-database queries and loadingHands-on
- Lakehouse or warehouse: choosing the right storeConcept
- Semantic models and Direct Lake modeHands-on
- DAX measures and star-schema design for reportsHands-on
- Building Power BI reports over Fabric dataHands-on
- Row-level and object-level security in modelsHands-on
A secure clinic operations report
Load curated appointment data into a warehouse, build a star-schema semantic model in Direct Lake mode and write five DAX measures, including no-show rate and average wait. Build a two-page report for clinic managers, add row-level security so each manager sees only their clinic, and test every role.
Knowledge checkpoint with explained answers
05Real-Time Intelligence and Governance
Build streaming analytics with eventstreams, eventhouses and KQL, and govern and deploy Fabric items properly.
- Real-Time Intelligence: eventstreams, eventhouses and KQL databasesConcept
- Ingesting streams with eventstreamsHands-on
- Querying with KQL and building real-time dashboardsHands-on
- Alerts and actions on live data with ActivatorHands-on
- Governance: sensitivity labels, lineage and the Purview hubHands-on
- Deployment pipelines and Git integration for Fabric itemsHands-on
Live waiting-room monitoring, governed
Stream synthetic clinic check-in events through an eventstream into an eventhouse and write KQL that computes live waiting times by clinic in five-minute windows. Build a real-time dashboard and a rule that alerts when any clinic’s wait exceeds a threshold. Label the data as confidential, show lineage to the dashboard and promote the items from development to test.
Knowledge checkpoint with explained answers
PART 3AI and Capstone
06Data Science, Copilot, Data Agents and Capstone
Train and apply models in Fabric, use Copilot and data agents responsibly, and plan the capstone.
- Data science in Fabric: notebooks, experiments and MLflowHands-on
- Training and scoring models with batch predictionHands-on
- Copilot in Fabric for notebooks, SQL, pipelines and reportsHands-on
- Fabric data agents: grounding natural-language answers in your dataHands-on
- Governing AI features: capacity, permissions and reviewConcept
- Capstone planning and reviewHands-on
Predict no-shows and ask the data
Train a model that predicts appointment no-shows from the gold tables, track at least three runs and score next week’s appointments in batch. Configure a data agent over the same tables that answers five manager questions, and check each answer against a query you write yourself. Use Copilot to draft one report page and record what you had to correct.
Knowledge checkpoint with explained answers
Wrenfield Community Clinics operations analytics in Fabric
Wrenfield Community Clinics is a fictional network of twelve outpatient clinics. You receive synthetic appointment extracts, a staff roster spreadsheet, a live stream of check-in events and a mirrored scheduling database. Build a Fabric solution that reports no-show rates and staffing against demand, monitors waiting times live, predicts next week’s no-shows and lets managers ask questions of their data through a verified data agent. All data and the organisation are synthetic, and no real patient data is involved.
Workspaces, capacity and governance — Set up workspaces, domains, roles, labels and Git integration before building.
Ingestion and lakehouse engineering — Land and refine every source in a medallion lakehouse.
Warehouse, semantic model and reports — Serve the questions through a secure Direct Lake model and reports.
Live waiting-time monitoring — Monitor waiting times as check-ins arrive and alert on breaches.
Prediction and data agent — Predict no-shows and give managers verified natural-language answers.
Next course
Tell us about yourself or your team and we’ll let you know how and when you can take this course.