Modelling and prediction of turbidity in water

Abstract

Clean water is vital to our health. For effective water quality management, fast and efficient alternative methods such as artificial intelligence (AI) models are needed. In this analysis, New Zealand (NZ) river water data (https://shiny.niwa.co.nz/nzrivermaps) is used. Spatial modelling of this dataset is done by the authors in (Whitehead, A, 2018). In current analysis, a knowledge driven machine learning based computational model for predicting turbidity measures is developed using Different machine learning (ML) Regressors such as Decision Tree (DTR), Support Vector (SVR), Random Forest (RFR) and Gradient Boosting (GBR). These algorithms are assessed and compared based on their performance metrics: Mean absolute error (MAE), Root Mean Square Error (RMSE) and coefficient of determination (Table1). Pipeline of the machine learning model is shown in Figure 1.