Comparison of two cluster analysis methods using single particle mass spectra

Cluster analysis of aerosol time-of-flight mass spectrometry (ATOFMS) data has been an effective tool for the identification of possible sources of ambient aerosols. In this study, the clustering results of two typical methods, adaptive resonance theory-based neural networks-2a (ART-2a) and density-...

Full description

Saved in:
Bibliographic Details
Published in:Atmospheric environment (1994) 2008-02, Vol.42 (5), p.881-892
Main Authors: Zhao, Weixiang, Hopke, Philip K., Prather, Kimberly A.
Format: Article
Language:English
Subjects:
Adaptive resonance theory neural networks
Aerosol time-of-flight mass spectrometer
Analysis methods
Applied sciences
Atmospheric pollution
Density-based cluster analysis
Exact sciences and technology
Pollution
Single particle
Source identification
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Cluster analysis of aerosol time-of-flight mass spectrometry (ATOFMS) data has been an effective tool for the identification of possible sources of ambient aerosols. In this study, the clustering results of two typical methods, adaptive resonance theory-based neural networks-2a (ART-2a) and density-based clustering of application with noise (DBSCAN), on ATOFMS data were investigated by employing a set of benchmark ATOFMS data. The advantages and disadvantages of these two methods are discussed and some feasible remedies proposed for problems encountered in the clustering process. The results of this study will provide promising directions for future work on ambient aerosol cluster analysis, suggesting a more effective and feasible clustering strategy based on the integration of ART-2a and DBSCAN.
ISSN:1352-2310
1873-2844
DOI:10.1016/j.atmosenv.2007.10.024