Real-world datasets built by a BI architect. Fact and dimension tables separated, a data dictionary in every set, and a deliberately broken copy of each one. No Kaggle toy data — this looks like what you actually get after a CRM export.
| domain | contents | variants | format | status |
|---|
Structured around real business domains — sales pipelines, finance P&Ls, HR headcount — not abstract arrays. The kind of data you would actually find after a CRM export.
Every set ships twice: a model-ready version, and a version with duplicates, nulls, formatting drift, and broken relationships. Practice the whole workflow, not just the fun part.
Fact and dimension tables separated, with a data dictionary included. Built for practicing star schemas, DAX measures, and semantic layer design — not just pivot tables.
| status | release | what shipped |
|---|---|---|
| live | Sales & Revenue | 200-row opportunity fact table, 5 dimension tables, clean and dirty variants, full data dictionary. |
| live | Marketing + HR | 400-row lead funnel, weekly campaign performance, 150-employee headcount snapshots, and hiring pipeline data. Clean and dirty variants throughout. |
| live | Finance & FP&A | 336-row GL transaction fact table, quarterly budget entries, and 5 dimension tables covering accounts, departments, and fiscal periods. |
| live | Dirty Data Challenges | Five standalone challenge files — duplicates, orphaned keys, date format chaos, null patterns (MCAR/MAR/MNAR), and type chaos — each with a documented answer key. |
| live | Star Schema Kits | Retail and e-commerce kit: 2,000-row fact_sales, a date dimension with 18 attributes including fiscal calendar and holiday flags, 5 dimension tables, a relationship map, and ready-to-paste DAX and SQL. |