Gradient Boosting Model for Water Quality Forecasting

The aim of our study is to evaluate the quality of water using machine learning methods. Potability is a subjective word that quantifies water quality. In this research, a variety of metrics were used to measure general potability of the water. The prescribed elements are applied like variable matrix to denote water quality. The article used three different classification methods, with success rates ranging from 84.27 to 93.16%, to predict the water quality class including ensemble approach like Random Forest (RFC), SVM for classification (SVC), as well as XGBoost. The Gradient Boosting approach, out of all the ones used, received the greatest accuracy (93.16%), precision (96.3%), recall (92.76%) and F-score (94.54%). Also, it achieved the least computational delay with a mean latency of 1.34 s.