AI Based Ground Water Quality Prediction in Gujarat Province

This study emphases on predicting groundwater quality using Weighted Arithmetic Water Quality Index (WAWQI), Entropy-based Water Quality Index (EBWQI), and Canadian Council of Ministers of Environment Water Quality Index (CCME-WQI) methods, alongside machine learning models. Data from the Central Ground Water Board, Ministry of Water Resources, Government of India, are analyzed after extraction from PDF format. Various regression models, including Linear Regression, XGBRegressor, LightGBM Regressor, Catboost Regressor, and Random Forest Regressor, are employed for WQI prediction, with stratified k-fold vali-dation addressing class imbalance. Results highlight the CatBoost model’s superior performance with R2-scores of 0.9943 for WAWQI, 0.9934 for EBWQI, and 0.9918 for CCME-WQI. While Random Forest performed best with 0.9825 for WAWQI. And XGBoost performed best with 0.9932 for WAWQI. Identified factors influencing groundwater quality provide insights for improved water resource management strategies. The formulated predictive models provide a reliable way for assessment and monitoring of quality of groundwater, with potential applications of environmental planning and decision-making processes.