Prediction of River Water Quality Index and Uncertainty Analysis using Machine Learning Techniques

The increasing demand for clean water in river systems drives the development of new models for water quality prediction. The problem statement is the need for accurate and reliable prediction of water quality indices in various river basins using machine learning techniques. Through the integration of diverse methodologies in the proposed approach aims to overcome limitations associated with conventional methods and emerging in water quality concerns. The dataset used in Water Quality Prediction (WQP) in the river dataset was used to predict the water quality index. The proposed method in this paper consists of preprocessing, data standardization, and data splitting with the parameter and calculating the WQI. The combined machine learning algorithms include Random Forest (RF), Extreme Gradient Boosting (XGB), and K-Nearest Neighbor (KNN). The numerical validation of this paper is an MSE of 0.03, an RMSE of 0.09, an MAE of 0.02, and $R^{2}~\mathbf{0.12}$. The existing methods like Water Quality Management Limitations (WQM-WQI), Sequential Minimal Optimization-Support Vector Machine (SMO-SVM), Principal Quality Analysis (PQA-WQI), and Artificial Neural Network (ANN-WQI) are overcome by the proposed method.