Evaluation of Machine Learning Algorithm on Drinking Water Quality for Better Sustainability

Abstract

Water has become intricately linked to the United Nations’ sixteen sustainable development goals. Access to clean drinking water is crucial for health, a fundamental human right, and a component of successful health protection policies. Clean water is a significant health and development issue on a national, regional, and local level. Investments in water supply and sanitation have been shown to produce a net economic advantage in some areas because they reduce adverse health effects and medical expenses more than they cost to implement. However, numerous pollutants are affecting the quality of drinking water. This study evaluates the efficiency of using machine learning (ML) techniques in order to predict the quality of water. Thus, in this paper, a machine learning classifier model is built to predict the quality of water using a real dataset. First, significant features are selected. In the case of the used dataset, all measured characteristics are chosen. Data are split into training and testing subsets. A set of existing ML algorithms is applied, and the results are compared in terms of precision, recall, F1 score, and ROC curve. The results show that support vector machine and k-nearest neighbor are better according to F1-score and ROC AUC values. However, The LASSO LARS and stochastic gradient descent are better based on recall values.

Prediction of drinking water quality with machine learning models: A public health nursing approach

“…It was stated that KNN in terms of precision, LASSO LARS (LL), Stochastic Gradient Descent (SGD) in terms of recall, SVM, and Artificial Neural Network (ANN) in terms of ROC curve/AUC were good classifiers. Although XGB was used in the study, it was reported to give moderate results (Kaddoura, 2022). Nasir et al.…”

Impact of Public Service Quality on the Efficiency of the Water Industry: Evidence from 147 Cities in China

“…Second, it is necessary to further strictly control the maintenance of infrastructure, such as through the supervision of pipe networks and related technical indicators [36,69], to ensure that the water demand of the vast number of users is met. Third, attention should be paid to improving the efficiency of capital use and not blindly investing too much money in maintaining the pipe network infrastructure because there is no obvious relationship between the qualified rate of pipe network pressure and scale efficiency [70]. The water industry cannot expand its scale without limit but should control the scale of expansion within a reasonable range and focus on improving management efficiency.…”

Improving the Water Quality Classification Model for Various Farms Using Features Based on Artificial Neural Network

“…Twelve MLPNN-based models were evaluated for performance in the training and validation process with accuracy (Acc), precision (Prec), sensitivity (Sens), F1-score, and MAE. These efficiencies were formulated from ( 16) to (19) [38], [39], [40].…”

“…During the training process, the categorical crossentropy loss was used to evaluate the model's efficiency for multi-class in neural networks, which is formulated in (20) [40].…”