Data Analytics Academy

Day 4 of 5 · 4 hours

Power BI — interactive dashboards {: .hero-day__title }

Wrap the analysis in something a non-analyst will actually click through. Model, measure, visualise, ship.

Duration: 4 hours Prerequisites: Days 1–3 Learning goals: by end of day you can import data into Power BI, set up relationships across tables, write the 4 most useful DAX measures, and build a 1-page interactive dashboard that a non-analyst can use without you in the room.

Why Power BI after pandas?

Pandas + matplotlib can make any chart, but the audience for charts isn’t analysts — it’s executives, marketers, ops teams. Power BI’s job is to wrap your analysis in something they’ll actually click through. A good dashboard answers tomorrow’s question, not just today’s.

Tooling

  • Power BI Desktop — free, Windows only. Download: powerbi.microsoft.com/desktop.
  • Mac users: use a Windows VM, RDP into a Windows machine, or use the Power BI Service (browser) — Service has fewer features but works.
  • File extension: .pbix (one file = data model + visuals + measures).

The three apps inside Power BI

Pane / viewJob
Power QueryLoad-time transformations: rename columns, change types, derive fields. The “ETL” step. (Lesson 1)
Model viewTables and the relationships between them. (Lesson 2)
Report viewThe dashboard canvas. Visuals, slicers, layout. (Lessons 3 & 5)
DAXThe formula language for measures and calculated columns. (Lesson 4)

You’ll be switching between these all day. Knowing which one to reach for is half of Power BI competence.

Agenda

TimeBlockTopicLesson
00:00–00:30Hour 1Get Data & Power QueryLesson 1
00:30–00:55Hour 1The data modelLesson 2
00:55–01:05Break
01:05–01:35Hour 2Building visualsLesson 3
01:35–02:05Hour 2DAX measuresLesson 4
02:05–02:15Break
02:15–02:40Hour 3Interactivity, layout, polishLesson 5
02:40–02:55Self-test12-question gate-checkSelf-test
02:55–03:00Break
03:00–04:00Hour 4Capstone — build the dashboardCapstone

What you’ll be able to do

By the end of today, given a folder of CSVs and a business question, you can:

  • Import them into Power BI with clean column types and friendly names
  • Build a correct data model with one-to-many relationships
  • Pick the right visual for each part of the question
  • Write the four DAX functions that power 80% of dashboards
  • Add interactivity (slicers, cross-filters) and lay out a single-page report
  • Export it as a PNG / PDF for embedding or email

Lessons

Practice is folded into each lesson as collapsible “Try it yourself” boxes — read the concept, attempt in Power BI, reveal the sanity checks.

Self-test

When you’ve worked through all five lessons, take the 12-question self-test before the capstone. ~15 minutes.

Capstone task for today

../../capstone/day4_powerbi/README.md — build the Seller & Satisfaction Dashboard: 3 KPI cards, a category bar chart, a Brazilian state map, a bottom-10-sellers table, and three slicers.

Common pitfalls

  • Wrong relationship cardinality. Many-to-many *─* “works” but inflates aggregates. Confirm 1 ─ * in Model view.
  • DAX in Power Query, Power Query in DAX. Each tool has its job. Derived columns at load time → Power Query. Aggregates that depend on user filters → DAX measures.
  • Implicit measures everywhere. “Sum of price” is anonymous; you can’t reuse it. Create explicit named measures.
  • Visual clutter. A good 1-page dashboard has 4–6 visuals. Not 12.
  • Default aggregation = Sum. Sum of star ratings is meaningless. Check every visual’s aggregation explicitly.