Ozone Concentration Forecasting: Assessing the Efficacy of MLP, DNN, and XGBoost in Environmental Bench-AQ Dataset

This study investigates the prediction of tropospheric ozone concentration, a critical aspect of air quality assessment. We utilized machine learning techniques, including Multilayer Perceptron (MLP), Deep Neural Network (DNN), and XGBoost, and analyzed extensive air quality datasets that incorporated ozone-specific metrics and environmental variables. The findings revealed that XGBoost notably excelled over MLP and DNN in predicting average ozone concentrations and Aot40 metrics, as evidenced by its lower Mean Absolute Error (MAE) values. This research highlights the importance of precise feature selection in machine learning and demonstrates the potential of advanced ensemble methods like XGBoost in environmental modeling. It suggests the need for future exploration in integrating a broader range of environmental variables and applying these findings in real-world scenarios, especially in urban planning and public health.