Clustering on Air Quality Data in the United States

Abstract

The grouping of the temporal characteristics of air quality pattern is worth taking into account when investigating the air quality situation in an area. In this project, Euclidean k-means, Soft-DTW-k-means and Self-Organizing Map(SOM) are used and compared for the exploration of clustering on time series air quality data in the Continental United States. In the results, the concentration of PM2.5 and SO2 in 2019 did not show distinct spatial clusters while NO2 had an apparent cluster in the California and O3 also had different seasonal fluctuation patterns in corresponding regions.