Data analysis and preprocessing techniques for air quality prediction: a survey

• Published in: Stochastic environmental research and risk assessment 2024-06, Vol.38 (6), p.2095-2117
• Main Authors: Yu, Chengqing, Tan, Jing, Cheng, Yihan, Mi, Xiwei
• Format: Article
• Language: English
• Subjects: Air pollution; Air quality; Algorithms; Aquatic Pollution; Chemistry and Earth Sciences; Cognition & reasoning; Computational Intelligence; Computer Science; Data analysis; Data processing; Decomposition; Earth and Environmental Science; Earth Sciences; Emergency communications systems; Environment; Environmental governance; Environmental research; Feature selection; Forecasting; Interpolation; Math. Appl. in Environmental Science; Missing data; Nitrogen dioxide; Nonlinear systems; Outdoor air quality; Outliers (statistics); Physics; Pollutants; Preprocessing; Probability Theory and Stochastic Processes; Review Paper; Risk assessment; Spatial data; Statistics for Engineering; Waste Water Technology; Water Management; Water Pollution Control
• ISSN: 1436-3240; 1436-3259
• DOI: 10.1007/s00477-024-02693-4

Air quality prediction technology can provide effective technical means for environmental governance. In recent years, due to the strong nonlinearity of data, there has been extensive research on data analysis and preprocessing techniques. This paper aims to comprehensively summarize and analyze the methods used in air quality forecasting, specifically focusing on four categories: data decomposition, dimensionality reduction, data correction, and spatial interpolation. Each method's purpose, characteristics, improvements, and implementation details are described in detail. The evaluation of data preprocessing methods is based on popularity, accuracy improvements, time consumption, maturity, and implementation difficulty. Among the existing methods, data decomposition and feature selection are commonly used and well-developed. However, outlier detection and spatial interpolation have limited applications and require further research. Furthermore, this paper discusses current challenges in applying these methods and future development trends, providing a valuable reference for future research.