MACHINE LEARNING FOR PREDICTING DRINKING WATER QUALITY

Abstract

Water conservation and care is one of the most important environmental problems today. Water quality refers to the appropriate values ​​of the physicochemical and / or biological parameters of the water for a specific use and its monitoring provides useful information to be processed by machine learning tools for predictive purposes. This document aims to present a review of machine learning techniques used in estimating water quality. Research works show that neural networks (RN), neuro diffuse inference systems (ANFIS), and support vector machines (MVS) are the most widely used predictive techniques, the results obtained in the accuracy measures show the viability of estimate the quality of water in rivers, basins, lakes, among others. Comparative studies propose new strategies to improve the accuracy of the predictions, which integrates the properties of two or more techniques to form hybrid models, their implementation with different algorithms and methodologies can be extensive and is emerging as a future scenario to be developed.