Comparative Study of Machine Learning Models for Real-Time River Pollution Monitoring and Prediction Based on Key Water Quality Parameters in the Cuttack City, Odisha

River pollution has emerged as a critical environmental issue worldwide, affecting water quality, aquatic ecosystems, and human health. Timely and accurate detection of pollution is essential for effective water resource management and environmental protection. This study investigates the application of machine learning (ML) algorithms for real-time monitoring and prediction of river pollution using key water quality parameters, including temperature, pH, turbidity, electrical conductivity (EC), and dissolved oxygen (DO). Three ML models such as: Decision Tree (DT), Support Vector Machine (SVM), and Neural Network (NN), were employed to investigate continuous sensor data collected from the Mahanadi River in Cuttack City, Odisha. The performance of the models was evaluated using accuracy, precision, recall, and F1-score, and the relative importance of each water quality parameter was assessed to identify critical contributors to pollution detection. Findings indicate that Neural Networks outperformed DT and SVM, achieving an accuracy of 92