Enhancing surface water quality prediction efficiency in northeastern thailand using machine learning

Abstract

Water is the most vital resource for life and is necessary for most living creatures, including humans, to survive. Three rivers’ surface water quality has been predicted by this study: the Chi river, the Mun river, and the Songkhram river. In the northeastern region of Thailand. The dataset is 881 samples and 13 factors. This study investigated various machine learning methods for predicting water quality, including neural networks (NN), support vector machines (SVM), decision trees (DT), Naive Bayes (NB), and K-nearest neighbors (KNN). Furthermore, this study was conducted to find suitable factors using correlation based feature selection, correlation coefficient, and information gain. And optimize the prediction model using the Bagging Approach. The result is found that the bagging model using the DT technique (BaggingDT) has better performance than all models with an accuracy value equal to 98.64%, precision value equal to 98.70%, recall value equal to 98.60%, F-measure value equal to 98.60% and RMSE value equal to 0.0961. The obtained factors and the most appropriate model can be used to develop a surface water quality standard predicting system.