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

- Schema: observation_date (YYYY-MM-DD, weekdays only, sorted, no duplicates), DGS10 (percent, 2 decimals).
- Range 2000-01-03 … 2019-12-31. Weekday counts of rows: Mon 1044, Tue 1044, Wed 1043, Thu 1043, Fri 1043.
- Empty DGS10: 215 rows (US market holidays, incl. 2001-09-11/12 closure). No other non-numeric values.
- Available yields: 5002, range 1.37–6.79 %, plausible.
- Calendar gaps between consecutive available obs: 1 day 3899, 2 days 58, 3 days 889, 4 days 155. Changes after a holiday span >1 trading day; per the task rules they are kept and attributed to the weekday of t (documented as a data issue).
- No input violates availability rules; the train/test split is enforced in code.