A Review on Air Pollution Prediction Using Artificial Intelligence

Air pollution has become a risk to the ecosystem in recent years, raising health concerns. To take precautionary steps against air pollution, policymakers require precise pollution forecasts achieved through advanced techniques. This paper conducted an extensive literature review of machine learning and deep learning models used for air pollution forecasting. The literature review consisted of 60 studies, each of which focused on indoor and outdoor air pollution forecast models. In comparison, hybrid and ensemble models emerged as prominent models across multiple regions using datasets collected from monitoring stations, satellites, and sensors. A few articles included interpretable models to evaluate the correlation between the target feature and the features used for model predictions through Explainable Artificial Intelligence techniques. This study offers a comparative discussion and further enhancements to existing studies, which can help improve performance and interpretability. The use of deep learning models was highlighted for their consistent performance in spatiotemporal modelling, along with various suggestions for future enhancements.