Classification of Quality of Water Using Machine Learning

Abstract

Keeping a check on the quality of water is necessary for protecting both the health of humans and of the environment. AI can be used to classify and predict water quality. The proposed system uses several machine learning algorithms to manage water quality data gathered over a protracted period. Water quality index (WQI) is used to categorize the given samples by using machine learning and ensemble approaches. The studied classifiers included random forest classifier, CatBoost classifier, k nearest neighbors, logistic regression, etc. The authors used precision-recall curves, ROC curves and confusion matrices as performance metrics for the ML classifiers used. With an accuracy of 95.43%, the random forest model was shown to be the most accurate classifier. Furthermore, CatBoost classifier and k nearest neighbors provided satisfactory results with 94.86% and 94.08% accuracy, respectively. Therefore, CatBoost algorithm is considered to be a more reliable approach for the quality of water classification.