Answers are grounded in this paper and cite their sources.
Annual drinking-water forecasts support medium-term infrastructure planning with limited data. Using 2006–2023 data from Adana, Türkiye, this study combines Step-Wise Weight Assessment Ratio Analysis (SWARA), principal component analysis (PCA), and an Aydeniz drought coefficient. SWARA prioritized population, registered water subscribers, and urban development; five PCA components explained 88.75% of predictor variance. Population, current-price gross domestic product, subscribers, and drought stress informed Lasso, Ridge, XGBoost, and support vector regression (SVR). Models were developed on 2006–2020 data and externally evaluated on 2021–2023. SVR yielded the lowest normalized errors (MAE = 0.02775; RMSE = 0.03915) and a Nash–Sutcliffe efficiency (NSE) of 0.77878. Fifteen leave-one-development-year-out refits yielded NSE values of 0.754–0.885. Scenario forecasts indicate distribution volume rising from 110.907 million m3 (2023) to 120.186 million m3 (2026), equivalent to 25,422 m3/day additional average demand. This increment supports planning, not direct design capacity.
No takes yet. Share an insight, caveat, or question.
Baltacı et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: