Machine Learning for Soft Sensing in Water Quality Monitoring: A Survey and Tutorial

Water quality monitoring is essential for environmental protection, public health, and sustainable water resource management. However, many important water quality parameters remain difficult or expensive to measure continuously using conventional sensing techniques. In recent years, Machine Learning (ML)-based soft sensing has emerged as an effective alternative for estimating hard-to-measure variables using readily available environmental and sensor data. This paper presents a comprehensive survey and tutorial on ML-based soft sensing in water quality monitoring. First, the study reviews and classifies existing approaches, including linear regression, support vector machines, tree-based ensemble methods, artificial neural networks, and advanced deep learning architectures. Second, the paper discusses major challenges, emerging trends, and research gaps related to data quality, sensor drift, interpretability, transferability, uncertainty quantification, and real-world deployment. Third, a tutorial-style framework is presented covering data preprocessing, feature engineering, model selection, training, evaluation, explainability, and deployment considerations for practical soft sensing development. The review further highlights emerging directions involving explainable artificial intelligence, edge intelligence, digital twins, remote sensing integration, and intelligent cyber–physical monitoring systems. Overall, this study provides researchers and practitioners with a comprehensive reference for developing reliable, scalable, and interpretable ML-based soft-sensing systems for water-quality monitoring.