This paper demonstrates a cost-effective approach to air pollution source apportionment using low-cost sensors, Positive Matrix Factorization (PMF), and k-means clustering. The methods successfully identified and quantified major pollution sources at three sites: a construction site, a quarry, and a roadside location. The analysis considered meteorological conditions and temporal variations, providing crucial information for air quality management at significantly lower costs than traditional methods.

Dimitrios Bousiotis, Gordon Allison, David C. S. Beddows, Roy M. Harrison, Francis D. Pope