Water Quality Analysis through Machine Learning and Deep Learning in IoT Systems: A Systematic Review

Abstract

The ecological quality of water is crucial for the protection of the aquatic environment and human health, and is affected by natural factors and, to a greater extent, by pollution resulting from industrialization, agriculture, and urbanization. This article presents a systematic review on the application of Machine Learning and Deep Learning in water quality analysis. The aim is to evaluate the effectiveness of Machine Learning and Deep Learning in water quality analysis, identifying accurate and reliable methods to develop advanced tools that facilitate the monitoring and prediction of this vital resource, thus improving its management and conservation. The PRISMA method was used to gather 65 significant articles related to water quality. The results suggest that Machine Learning and Deep Learning are fundamental in this field, particularly in studies conducted in China and India. The most common algorithms in Machine Learning are Random Forest and SVM, while in Deep Learning LSTM and CNN stand out. It is concluded that Machine Learning and Deep Learning are essential to assess water quality with IoT, the choice between the two depends on the availability of data and the objectives of the analysis, Machine Learning is preferable with limited data and limited resources, while Deep Learning is more effective with large volumes of data.