Development of regression models for predicting Water Quality Index based on Dissolved Oxygen for river pollution assessment / Muhamad Irfan Ahmad Suhkri

Water is a critical resource in Malaysia, but river water quality has been significantly impacted by industrialization, urbanization, and rapid development. Traditional water quality monitoring methods are time-consuming and often result in delayed detection of pollution. This study addresses these challenges by developing regression models to predict the Water Quality Index (WQI) based on Dissolved Oxygen (DO) measurements, enabling timely and efficient river pollution assessment. Data were collected from Kaggle, consisting of 219 clean and polluted river water samples from spanning June to November 2023. After data categorization, only 44 of clean and polluted DO data were used respectively. The data were categorized into clean and polluted water conditions and validated through statistical analyses, including normality tests and error bar plots. Multiple regression techniques, such as Linear Regression and Robust Linear Regression, were implemented using MATLAB and Python. Among the models tested, MATLAB's Linear Regression achieved the highest R2 value of 0.95397 and the lowest RMSE of 7.2728, demonstrating superior performance. These findings highlight the potential of regression modelling as a reliable and proactive approach to water quality monitoring, supporting environmental authorities and stakeholders in safeguarding freshwater ecosystems and improving resource management.