Employing Explainable Artificial Intelligence (XAI) for Enhanced Water Quality Assessment

Maintaining the purity and quality of water is crucial for the well-being of numerous living organisms, as water consumption directly impacts physical health. Unfortunately, Earth is experiencing a concerning decline in water quality, resulting in a plethora of diseases and health deficiencies among individuals. One of the main challenges in water quality management is the presence of Total Dissolved Solids (TDS) & various other substances including potassium, sodium, chlorides, lead, nitrate, cadmium, arsenic, and other pollutants, further underscoring the need for comprehensive measurement, control, and monitoring of water quality. This disregard for the fundamental necessity of high-quality water has reached a critical juncture, necessitating global attention. Consequently, efforts are underway to raise awareness and conduct research aimed at making safe drinking water accessible to all. Explainable Artificial Intelligence (XAI) offers a solution by clarifying how machine learning models reach decisions, highlighting the necessity for transparent and comprehensible models to gain trust and support from stakeholders. Three popular machine learning models k-nearest neighbour (KNN), Random Forest and Extreme Gradient Boosting (XGBoost) are utilized for data classification, followed by the application of SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME) techniques to identify key characteristics and variables impacting water quality predictions. These insights can enhance decision-making and facilitate the development of effective strategies for managing and reducing water pollution.