Research on Functional Data Method Based on Air Quality Index of Sichuan Basin

Abstract

So far, there have been relatively few studies on air quality in the Sichuan Basin. This paper uses the functional data analysis method to study the air quality data of the Sichuan Basin. Firstly, the discrete air quality data of 21 prefecture-level cities in the Sichuan Basin were smoothed using the Fourier basis function to transform them into continuous functional data. Then, the principal component analysis of the smoothed air quality index was conducted. Secondly, the non-negative matrix factorization method is introduced to achieve data dimensionality reduction and retain key information through the linear combination of basis vectors, and to conduct cluster analysis on the air quality data of the Sichuan Basin. The research results show that the fluctuation of the air quality index in the Sichuan Basin has a distinct seasonal pattern. The cumulative variance contribution rate of the first three principal components reached 95.3%, which can explain most of the original data. According to the clustering results, the Sichuan Basin can be divided into three types of regions, namely the core area, the marginal area and the mountainous area in southwest China. These studies provide data support and theoretical basis for the future pollution prevention and control policies in the Sichuan Basin.