Water quality assessment in a dense urban landscape using Statistical Analysis (SA), Artificial Intelligence (AI) and Geographic Information System (GIS).

Rivers are vital freshwater resources, yet they are increasingly threatened by pollution from multiple sources in densely populated cities such as Dhaka, Bangladesh. These contaminants threaten environmental sustainability and public health, underscoring the challenges posed by urbanization. This study aims to develop a predictive model and identify the critical parameters that significantly influence the Water Quality Index (WQI) using Statistical Analysis (SA), Machine Learning (ML), and Geographic Information Systems (GIS). A total of 648 datasets from the Turag River in Dhaka were gathered, comprising sixteen features, including physicochemical and environmental variables. Descriptive Statistics such as Correlation, Regression, and ANOVA, were performed to assess correlations and significance. SA findings demonstrated significant correlations between the WQI and Total Dissolved Solids (TDS), Electrical Conductivity (EC), and Turbidity, while regression analysis indicates moderate to strong relationships (e.g., TDS: R2 = 0.678, Turbidity: R2 = 0.821). Three ML models, including Random Forest (RF), Decision Tree (DT), and XGBoost, were employed to forecast WQI and identify significant determinants. All ML models achieved over 90% accuracy, demonstrating their applicability to real-world scenarios. Mean Squared Error (MSE) and Root Mean Squared Error (RMSE) were used to assess the model's performance. GIS was enabled tracking changes in water quality at specific sites over time, which helped identify monthly trends and pollution patterns. Thus, the study demonstrates the substantial influence of environmental and physicochemical factors on water quality and underscores the significance of AI-supported monitoring systems in enhancing public health, early warning capabilities, and pollution control.