Role of artificial intelligence and learning techniques in time series forecast of PM2.5 data: An extensive review

Nowadays, this survey is necessary to equip the society with enriched Particulate Matter 2.5 (PM2.5) forecasting and enabling the health interventions with decision making policy. For researchers and academicians, this review provides an opportunity to explore the state-of-art about the role of Artificial Intelligence (AI) and Machine Learning (ML) techniques in the process of predicting the future air quality by analyzing the time series air quality data. For this purpose, the various prediction systems have been developed by various researchers in the past for predicting the various life-threatening diseases, natural calamity and air pollution by using statistical approach, ML and Deep Learning (DL) techniques. This article discusses about the various time series forecasting techniques that are developed by using Statistical approach, ML and DL Techniques. Especially, the time series forecasting methods on PM2.5 data are considered in this review by highlighting the method differences and the result comparative analysis. The PM2.5 data is widely used dataset to predict the future air pollution level in the specific regions of the world. In this review, the number of research papers published in these areas. This review shows the various experimental results of the different research works are summarized with respect to the evaluation metrics including Mean Absolute Error and prediction accuracy. Finally, a new idea is proposed in the suggestion proposed section by highlighting the merits.