Demographic and life-table forecast dataset for Uzbekistan, 2020–2050: a hybrid Lee–Carter–Gompertz mortality-modelling approach

This dataset provides abridged life tables and mortality indicators for Uzbekistan at the national level, covering historical reference years (2000, 2005, 2010, 2015, 2019) and annual forecasts for 2020–2050. Seven life-table functions are reported — central death rate (nMx), probability of dying (nqx), survivors of a radix of 100,000 (lx), deaths (ndx), person-years lived (nLx), total person-years remaining (Tx) and life expectancy (ex) — across 19 abridged age groups (<1, 1–4, 5–9, …, 80–84, 85+). Historical indicators are given for both sexes, males and females; forecasts are given for males and females. Historical rates are drawn from the WHO Global Health Observatory. The 2008 all-population empirical life table (single year of age) is from the State Committee of the Republic of Uzbekistan on Statistics (www.stat.uz). Forecasts were produced with a hybrid Lee–Carter–Gompertz model in which the log central death rate is expressed as ln m_x(t) = (ln A + B·x) + a_x + b_x·k_t + ε, with the mortality index k_t extrapolated by an ARIMA(1,1,1) process. Data are provided in tidy (long) CSV format keyed by year × sex × age group × indicator, together with a multi-sheet Excel workbook, two single-year reference life tables for 2008 (empirical and Gompertz), a model-comparison table (R²/RMSE by model and sex), a data dictionary, and a reproducible Python build script. Each value carries a provenance flag: "published" values are transcribed directly from the source, while "derived" forecast values (ndx, nLx, Tx) are reconstructed from published lx using the standard abridged life-table algorithm. All life-table identities were validated programmatically (0 ≤ nqx ≤ 1; monotone lx; ndx = lx(x) − lx(x+n); ex = Tx/lx); a validation report is included.