Data audit
| file | SHA-256 | rows (excl. header) | cols |
|---|---|---|---|
| data/DGS10.csv | 8187f89ab6523b3c964255e1736cfcf0e2ca854ed9e50343e04f4dc2945c5cd7 (matches README) | 5217 | 2 |
| data/README.md | provenance text only | – | – |
- Schema:
observation_date(YYYY-MM-DD),DGS10(percent, 2 decimals). Date range 2000-01-03 … 2019-12-31. - Duplicated dates: 0. Weekend dates: 0. Non-numeric non-empty values: 0.
- Empty DGS10: 215 rows (US market holidays) → dropped per task rule (exclusions.csv step 1).
- Yield range 1.37–6.79 %. Δy (bp): sd 5.72, min −51 (2009-03-18, FOMC QE announcement – genuine), max +25 (2001-11-15). No anomalies requiring removal.
- Changes after a holiday span >1 weekday (155 changes with gap > 3 calendar days); kept per task rule, sensitivity reported.
- Info-availability: single permitted input; no violations. Forecast weekday means use only changes dated ≤ 2009-12-31 (last train date 2009-12-31, first test date 2010-01-04).