Data audit
| file | SHA-256 | rows×cols | range | missing | dups |
|---|---|---|---|---|---|
| cpi_quarterly.csv | c8652dec1567ec35e3b84ee8fa55545e9bf0ae4990a4d6d05b89a6d22f6801ea (matches README) | 127×3 | 1995-01-01…2026-07-01 | infl_annualised empty for 1995Q1 only (by construction) | 0 |
| cpi_monthly.csv | 855d069176369d30d02cbbe0cc6456b76d2f3967cdaabc1b362c2461ef9015a2 (matches README) | 381×2 | 1995-01…2026-09 | 0 | 0 |
- Units: CPI index 2002=100 NSA; infl_annualised in % at annual rate.
- Regular spacing: all quarterly gaps 89–92 days, monthly 28–31 days: no gaps.
- Consistency: quarterly averages recomputed from monthly match to 3.3e-7; inflation recomputed from cpi_avg matches to 3.3e-6 (rounding of cpi_avg to 6 dp) — negligible; file values used as given.
- Inflation: mean 2.13, sd 2.39, min −6.04 (2020Q2 COVID), max 12.15 (2022 surge). Strong seasonality: Q2 mean 4.18 vs Q4 mean 0.30.
- Anomalies: COVID 2020 drop and 2021–22 surge are genuine; retained (no outlier treatment).
- Availability: the data run to 2026Q3; observations after an origin are used only as targets, enforced in code by slicing quarter ≤ origin. Monthly file is provenance only.