Deeper into Air Pollution Forecasting: A Comparative Study of Deep Learning Models on Air Pollution Forecasting

Abstract

Air pollution is a matter high concerns for urban areas. Outdoor air pollution is at a level which can seriously threaten and harm the human health and life in major cities, especially to elderly and children. Therefore, many countries in the world have constructed stations for monitoring air pollution around major cities to observe air pollutants such as PM 2.5, PM 10, CO, NO2, SO2 and to alert their citizens if there is a pollution index which excesses the quality threshold. Also, air pollution is impacted by the meteorological factors of local place such as temperature, humidity, rain, wind, etc. This research aims in comparative study to forecast the reading of air pollution by using various deep learning models. Finally, it does the investigation on accuracies of various models in predicting air-pollution and discussion on draw backs of models used. Through this project, there will be a significant motivation for not only continuing research on urban air quality but also help the government leverage these insights to enact beneficial policies.