Enhancing Water Sustainability with AI methods: Analysis and Prediction of Seasonal Water Quality of Nepal Using Machine Learning Approach: Prediction of Seasonal Water Quality of Nepal Using Machine Learning Approach

Biplov Paneru (Department of Electronics and Communication Engineering, Pokhara University, Bhaktapur Nepal)
Bishwash Paneru (Department of Applied Sciences and Chemical Engineering, Tribhuvan University, Lalitpur, Nepal)
Sanjog Chhetri Sapkota (Nepal Research and collaboration centre)

International Journal of Informatics, Information System and Computer Engineering (INJIISCOM)
Vol. 6 No. 1 (2025): INJIISCOM: VOLUME 6, ISSUE 1, JUNE 2025

Water quality monitoring is crucial globally, and particularly in Nepal, for ensuring safe drinking water and preventing waterborne illnesses. This study employs machine learning to analyze and forecast the seasonal Water Quality Index (WQI) of Nepalese well water. Hybrid models incorporating nested cross-validation were developed, evaluating algorithms such as CatBoost, Decision Trees, Logistic Regression, MLP-GRU, and LSTM-GRU. CatBoost achieved the highest classification accuracy (99.35%), while the LSTM-GRU hybrid excelled in capturing complex temporal patterns. Nested cross-validation demonstrated a robust 96.13% mean accuracy with low variance. Furthermore, SHAP analysis applied to a Support Vector Machine (SVM) model identified critical predictive factors. This research highlights the significant potential of machine learning in effectively managing and predicting water sustainability.