# Nadia's Q3 Revenue by Region — Schema

Flat CSV in `nadia-q3-region.csv`. Two logical row types in one file, disambiguated by the `row_type` column.

## Columns

| Column | Type | Description |
| --- | --- | --- |
| `row_type` | text | Either `quarter` (aggregate row) or `region` (per-region row). |
| `region` | text | `TOTAL` for the aggregate row; region code (`APAC`, `EMEA`, `LATAM`, `NA`) for regional rows. |
| `revenue_m` | decimal | Actual revenue in millions of USD. |
| `plan_m` | decimal | Plan revenue in millions of USD. |
| `variance_pct` | decimal | `(revenue_m - plan_m) / plan_m * 100`, rounded to one decimal. `0.0` for the quarter total; per-region for regional rows. |

## Row inventory

- 1 `quarter` row (`region = TOTAL`).
- 4 `region` rows (`APAC`, `EMEA`, `LATAM`, `NA`).

Total: 5 rows.

## Deliberate design choices

The dataset carries both actual and plan on every row so the reader can reproduce either the Copilot first-draft chart (which uses only `revenue_m`) or the hand-built chart (which needs both to compute per-region variance). The quarter row is included so any tool reading the CSV can confirm the regional rollup sums to the quarter total without cross-referencing a separate file.

The `variance_pct` column is pre-computed to fix the exact percentages Chapter 7's caption cites (`NA -5.3%`, `APAC +4.3%`, `EMEA +5.1%`, `LATAM +4.0%`). Rounding is done once at dataset build time so any downstream chart that displays these percentages matches the print figure exactly.

## License

Released for teaching use alongside *The Defensible Decision*. Use freely.