Data audit (P1)
| file | SHA-256 | rows (excl. header) | cols |
|---|---|---|---|
| data/cpi_quarterly.csv | c8652dec1567ec35e3b84ee8fa55545e9bf0ae4990a4d6d05b89a6d22f6801ea (matches README & bundle.json) | 127 | 3 (quarter, cpi_avg, infl_annualised) |
| data/cpi_monthly.csv | 855d069176369d30d02cbbe0cc6456b76d2f3967cdaabc1b362c2461ef9015a2 (matches) | 381 | 2 (month, cpi) |
- Quarterly: 1995-01-01 … 2026-07-01, 127 consecutive quarters, no duplicates, no gaps. Units: cpi_avg index 2002=100; infl_annualised % annual rate.
- Missing: infl_annualised empty for 1995Q1 only (by construction). Monthly: no missing, no duplicates, 1995-01 … 2026-09 consecutive.
- Consistency: quarterly cpi_avg recomputed from monthly matches to max abs diff 3.3e-7; infl_annualised recomputed from cpi_avg matches to 3.3e-6 (rounding).
- Distribution of infl_annualised: mean 2.13, sd 2.39, min −6.04 (2008Q4), max 12.15 (2022Q2, only |x|>10). Strong seasonality: quarter means Q1 2.32, Q2 4.18, Q3 1.66, Q4 0.30.
- Anomalies: none erroneous; large values (2008Q4 oil collapse, 2020Q2 COVID, 2022 surge) are genuine and kept.
- Availability: files contain data up to 2026Q3 (beyond last origin 2025Q3). Handling: analysis truncates the sample to quarters ≤ origin for every forecast; later quarters used only as realised targets (last target 2025Q3+4 = 2026-07-01, present).