Python‐Driven Multivariate Analysis: A Comprehensive Approach to Sustainable Groundwater Quality Assessment for Irrigation Purposes

ABSTRACT This study investigates groundwater quality in the Northern Ranebennur taluk of Haveri district, Karnataka, focusing on the geological, hydrological, and anthropogenic factors affecting water suitability for agriculture. A total of 150 samples from 50 villages were analyzed for parameters such as pH, electrical conductivity (EC), total hardness (TH), and sodium content. Alongside hydrochemical assessment, a bibliometric analysis was conducted to map global and regional research trends on groundwater quality and management, and this was further combined with machine learning approaches to provide both a literature‐driven perspective and a predictive framework for irrigation water suitability. Results showed a pH range of 6.6–8.2 and average EC values of 3.30 dS/m, indicating varied salinity. Sodium absorption ratio (SAR) ranged from 4.74 to 24.30, highlighting issues with soil permeability. While TH was classified as soft (6.71–62.00 mg/L), samples varied from fresh to brackish water (TDS: 435.20–3628.80 mg/L). Despite favorable conditions indicated by magnesium absorption ratio (MAR) and Kelley's Index, 91.33% of samples were only moderately suitable for irrigation, with many classified as “doubtful to unsuitable” based on the Wilcox and US Salinity Hazard diagrams. While predictive models, such as principal component regression (PCR), LASSO, Ridge Regression, and support vector machine regression (SVMR), performed well for most water quality indicators, including IWQ and TH, they still struggle to accurately predict Kelley's Index due to the complex interactions between ions. The study emphasizes the role of geochemical processes in groundwater quality and highlights the need for improved predictive modeling and groundwater management strategies to mitigate salinity and sodicity risks.