Prediction of water reuse potential in South African water management areas: A linear regression and Bayesian network approach

Abstract

ABSTRACT Increasing scarcity of freshwater resources and growing environmental awareness have given rise to the use of treated wastewater as an additional source of water supply. However, the amount of wastewater that can be reclaimed for reuse is subject to many factors, ranging from technical to socio-economic, environmental and institutional. This paper presents two prediction models for wastewater reuse potential in South Africa. The proposed models, namely a Bayesian Network model and a Linear Regression model, were used as a platform for integrated analysis of features such as water use, and socio-demographic and environmental factors. A dataset from the sub-regions of the 19 water management areas in South Africa was employed as input into the models in order to analyse these features and to consequently predict water reuse potential. Keywords: water reuse, wastewater recycling, Bayesian network, linear regression