Advanced Water Quality Assessment with IOT WSN and Multivariate Machine Learning Techniques

Ensuring clean and safe water is a critical global challenge, necessitating advanced monitoring techniques for early detection of contamination. This research presents an loT-driven water quality monitoring framework that integrates wireless sensor networks (WSNs) with multivariate machine learning models to assess key water parameters in real time. The proposed system collects and transmits data to a cloud platform for analysis, employing predictive algorithms to estimate E. coli levels. Six machine learning models-Ridge Regression, Random Forest Regressor, Stochastic Gradient Boosting, Support Vector Machine, k-Nearest Neighbors, and AdaBoost-were evaluated for their predictive accuracy. Among these, AdaBoost demonstrated the highest reliability, achieving the lowest mean absolute error (MAE) of 14.37 counts per 100 mL, while Stochastic Gradient Boosting showed the least accuracy. The study further identifies an optimal set of water quality parameters (Set A) comprising pH, conductivity, chloride, turbidity, nitrates, and chlorophyll, which enhances predictive efficiency. These findings highlight the potential of AI-driven water monitoring systems as early-warning tools, providing a cost-effective approach to ensuring water safety and public health.