# Data audit
| file | SHA-256 | rows (excl. header) | cols |
|---|---|---|---|
| data/DGS10.csv | 8187f89ab6523b3c964255e1736cfcf0e2ca854ed9e50343e04f4dc2945c5cd7 (matches README) | 5217 | 2 |
| data/README.md | 32b825f02387ef3d552df6def82c585d7f248980667a28f30e58506e472e561a | provenance text | – |

- Schema: observation_date (YYYY-MM-DD string), DGS10 (float, percent).
- Date range 2000-01-03 … 2019-12-31, strictly increasing, no duplicate dates, weekdays only (Mon 1044, Tue 1044, Wed 1043, Thu 1043, Fri 1043).
- Missing DGS10: 215 rows (holidays); no non-numeric non-empty values. Available obs 5002 → 5001 changes.
- Values 1.37–6.79 %, plausible; no anomalies.
- 214 changes span a holiday (calendar gap longer than normal); kept per TASK rule, flagged as a limitation.
- No information-availability violations in inputs; leakage risk only in forecasting step — handled by estimating weekday means on dates ≤ 2009-12-31 (last train date 2009-12-31, first test date 2010-01-04).