Intelligent modeling strategies for forecasting air quality time series: A review

• Published in: Applied soft computing Vol. 102; p. 106957
• Main Authors: Liu, Hui, Yan, Guangxi, Duan, Zhu, Chen, Chao
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.04.2021
• Subjects: Air quality forecasting; Hybrid modeling strategies; Intelligent predictors; Hybrid modeling strategies; Air quality forecasting; Intelligent predictors
• ISSN: 1568-4946; 1872-9681
• DOI: 10.1016/j.asoc.2020.106957

In recent years, the deterioration of air quality, the frequent events of the air contaminants, and the health impacts from that have caused continuous attention by the government and the public. Based on that, suitable and effective forecasting tools are urgently needed in scientific research. In this study, the basic forecasting algorithms are introduced as the simple forecasting models with their background, applications, advantages, and limitations, which include shallow predictors and deep learning predictors. Then, to enhance the forecasting ability, the data processing methods and two commonly used auxiliary methods (the ensemble learning and the metaheuristic optimization) in the hybrid models have been reviewed. The recent articles of the spatiotemporal aspects have also brought changes in both the analysis and the modeling methods. Furthermore, the representative models are summarized to present the structures of efficient predictive models. Some possible research directions of the air pollution forecasting are given at the end. This review aims to provide a comprehensive literature summary of the intelligent modeling strategies in the air quality forecasting, which may be helpful for subsequent study. •Intelligent models and the improved versions are reviewed.•Various components and combinations in the hybrid models are analyzed.•The applications of the forecasting models are provided and compared.•The future directions and challenges of air quality forecasting are discussed.