Use of Machine Learning for Realtime Water Quality Prediction

Abstract

Water is an essential resource that is a foundation not only for people’s lives but also for economic development. According to statistics from the World Health Organization (WHO) in 2021, over 2 billion people across the globe use water from contaminated sources leading to life-threatening diseases with over 485,000 deaths as a result of waterborne diseases being reported annually. There is therefore a need for tools and techniques that can be used to ensure water quality. Traditional water quality index calculations consume time and are often prone to errors. In the recent past use of Machine Learning (ML) in predicting water quality has generated interest among researchers as a real-time solution. From the reviewed literature, existing models only consider a few chemical properties of water while ignoring others. In addition, there is a need to compare the performance of different models. This study thus presents a supervised learning model for predicting water quality. Our hypotheses were that (i) most of such parameters are all important and none can be ignored and (ii) with supervised learning one can predict water quality based on its physio-chemical properties. Open data sets are explored as inputs with different models being evaluated. The results show that the Random Forest algorithm gives the best results with a 79 percent accuracy. The use of this model will go a long way towards ensuring water quality and thus reducing related diseases and fatalities.