Leveraging ML with XGBoost, CatBoost and LGBoost Classifiers to Optimize Water Quality Assessment and Prediction
Preet Singh, Taniya Hasija, K. R. Ramkumar
Abstract
Water is a key valuable resource for the survival of every living being. Water quality is an important concern at the global level as the pollution level continues to rise. Therefore, there is a need to develop methods to control and classify the water based on impurities present. The Machine Learning (ML) classification approach classifies the labels of the dataset for efficient categorical analysis. The Telangana state water quality dataset is taken for analysis purposes. The water quality is divided into 4 major classes. The three classification models used for this comparative analysis are XGBoost, LGBoost and CatBoost. According to the comparative analysis, the overall accuracy of CatBoost is higher out of all the classifiers used. However, the F1-score for the LGBoost classifier is higher for the water quality of class 1. This classification is essential in making smart decisions about sustainable water resource treatment and for taking action toward use to guarantee the quality of drinking water.