Practice — Mild · First Star

Skills focus. Loading the universe · the key check · creating relationships · proving filters flow.

Ungraded. This walks the assignment's road with the guardrails still up — do it first, and A04 becomes a rerun.

Tasks

Load sales.csv, returns.csv, products.csv, stores.csv, customers.csv. Checkpoint: 165,042 · 6,599 · 240 · 39 · 8,000 rows.
Before any relationships: run the key check on each dimension's ID column in the Power Query profile (or Table view). Checkpoint: products 240=240, stores 39=39, customers 8,000=8,000 — three clean keys. Then check sales.OrderID: 165,042 rows but 85,538 distinct. Not a key — a grain.
Open Model view. If auto-detect drew lines already, read each one; delete any you cannot explain. Create (or confirm) five: sales→products, sales→stores, sales→customers, returns→products, returns→stores. Checkpoint: every line reads *:1 with a single-direction arrow pointing at the fact.
Arrange the diagram as a star — facts center, dimensions around. Sixty seconds of dragging that pays off every time you open this file for the next three weeks.
Prove it on a report page:
- Card: sum of Quantity → 259,421
- Bar: count of sales rows by Category — Category from products → Lighting leads at 38,369 lines
- Slicer: Region from stores → clicking any region moves both visuals

Hint — my bar chart shows one bar labeled "blank"

The Category you dragged in cannot reach the rows you are counting — the relationship between that fact and products is missing, or you took Category from the wrong table. Model view first; check the line exists and reads *:1.

Reflection

Step 2 found three keys and one non-key. What would have happened if you had skipped it and tried to use sales as the "one" side of a relationship?
The bar chart mixes two tables — axis from products, values counted from sales. What piece of the model is doing the work that a merge did in Lesson 3?