A holistic data-centric water quality index model with endogenous aggregation weights for groundwater quality assessment: case of the Hodna basin in Algeria

This study proposes an integrated methodological framework combining Principal Component Analysis (PCA) and Data Envelopment Analysis (DEA) to construct a robust Water Quality Index (WQI) for groundwater assessment. Surrogate DEA inputs are generated by aggregating Optimistic Closeness Values (OCVs) derived from physicochemical parameters within each principal component, ensuring high information retention and early integration of benchmarking effects. Water quality classification is achieved through fuzzy J-means clustering, avoiding reliance on predefined thresholds. Spatial variability is examined through Kriging mapping, demonstrating the framework’s capacity to accurately capture groundwater quality patterns. Unlike expert-based indices prone to eclipsing problems, the proposed approach preserves parameter variability and critical information. Applied to 64 wells characterized by 15 parameters in Algeria’s Hodna Basin, the method classified 17.19% as Excellent, 10.9% Good, 18.75% Medium, 31.25% Marginal, and 21.87% Poor. Degraded quality is mainly attributed to gypsum presence and intensive agricultural practices, while superior quality dominates the Maadide region.