Predictive models of PM2.5 concentration with aerosol optical depth and meteorological data in Bangkok area using machine learning techniques

Air pollution, particularly fine particulate matter (PM2.5), is a significant global concern due to its adverse effects on human health and the environment. In response to this challenge, this study aimed to develop and compare machine learning models for predicting PM2.5 concentrations, focusing on two monitoring stations in Bangkok. A comprehensive dataset integrating meteorological data and aerosol optical depth (AOD) information was utilized. The models employed in this research included multiple linear regression (MLR), random forest (RF), and support vector machine (SVM). Notably, the SVM model demonstrated superior predictive performance for ambient stations. The findings underscore the importance of tailoring the machine learning method to the specific monitoring station type. Furthermore, the inclusion of influential gas variables such as NO2, SO2, CO, and O3 significantly enhanced the models' predictive capabilities. Fine-tuning hyperparameters further improved model performance. In conclusion, this research highlights the effectiveness of machine learning models in predicting PM2.5 concentrations, with important implications for air quality management in urban environments.