Forecasting water productivity in Somalia to 2032 using validated time-series models

This repository contains the data, complete R analysis code, numerical outputs, software session information, and figure-generating files supporting the study “Forecasting water productivity in Somalia to 2032 using validated time-series models.” The analysis uses 36 annual observations of total water productivity in Somalia from 1987 to 2022. Nine forecasting approaches were evaluated using a fixed hold-out period and expanding-window rolling-origin validation at 1-, 3-, and 5-year horizons. The repository includes the extracted Somalia dataset, reproducible R code, model accuracy results, ARIMA model specification and coefficients, forecasts for 2023–2032, residual diagnostics, and publication-quality figures. The source indicator was obtained from Our World in Data and originates from Food and Agriculture Organization AQUASTAT and World Bank data.