# Data audit
- data/life_expectancy_panel.csv — SHA-256 b3f2f774ed6797034acf62853286ca37fc02334ba3ce70e745e4f3d5c533c3ac (matches bundle.json). 439 data rows, 4 columns (country_name str, iso3 str, year int, life_expectancy float, years). Years 2000–2022. No whitespace issues.
- data/README.md — SHA-256 e9099d20…d0 (matches bundle.json).
- Anomalies:
  1. TUR appears 46 times: 23 rows "Turkey" + 23 rows "Turkiye", values identical for every year → exact duplicates from the two-extraction assembly. Handling: keep one per (iso3, year).
  2. CHE 2007 life_expectancy empty (1 missing value in file) → CHE incomplete.
  3. BGD, ETH, IND, NGA have rows only for 2011–2022 (12 rows each) → incomplete (likely the narrower extraction).
- Value range 51.49–84.56 years; largest year-on-year moves are 2020–2022 (COVID era, e.g. IND +4.4 in 2022), plausible; no unit errors found.
- No input violates the information rules; nothing imputed.