CC-BY-NC-NDVoza, DanijelaDehghani, HesamVeličković, Milica2024-01-102024-01-102023978-86-6305-140-9https://ioc.tfbor.bg.ac.rs/public/2023/Proceedings_IOC_2023.pdfhttps://repozitorijum.tfbor.bg.ac.rs/handle/123456789/5794One of the most reliable indicators of surface water quality is dissolved oxygen (DO). In order to take timely reactions in water pollution prevention and reduction, it is very useful to predict the changes in this parameter. In this study, a comparative analysis of the efficiency of different machine learning models in DO prediction was carried out. The aim was to examine which of the selected techniques indicate the best performance in DO prediction - Decision Tree, Random Forest, Gradient Boost Regression, Support Vector Regression, Multi-Layer Perceptron, and K - Nearest Neighbors Regression (KNN). According to the results, it can be concluded the best-fitted model on the created dataset is KNN.enDissolved OxygenMachine learningPredictionWater qualityTisa RiverconferenceObject