Modelling Air Pollution in Thailand: Insights from Community Mobility Data

This research investigates the relationship between community mobility and air pollution in Thailand, utilizing econometric and machine learning approaches to provide useful insights for policymakers to counter this issue. Data was sourced from the pollution database provided by the Ministry of Natural Resources and Environment and the community mobility database from a Google Trend search. The methodology of the research includes data extracting and pre-processing. The data analysis used an econometric model utilized Generalized Method of Moments, and a Machine Learning employed Support Vector Machines Results of the econometric analysis reveal that residential mobility, workplace mobility, and park mobility have a significant positive relationship with changes in air pollution. The support vector machine results show that community mobility explains 58.50% of air pollution variation and has a prediction accuracy of 94.47% on the training set. The results also suggest that pollution problems should be monitored closely when air pollution changes by 20%. These findings enhance the understanding of the complex factors influencing air pollution and offer valuable insights for developing effective mitigation strategies.