Contemporary air quality forecasting methods: A comparative analysis between statistical methods and classification algorithms

By I.N. Athanasiadis, K. Karatzas & P.A. Mitkas
In 5th Int'l Conference on Urban Air Quality Measurement, Modelling and Management , (R. Sokhi & J. Brexhler, ed.) , Valencia, Spain, 2005.

Abstract:
Cover image Urban air quality management involves forecasting modules, which are an essential component of related decision making processes, especially in densely populated cities, such as the Greater Athens Area (GAA). As many forecasting methods are being proposed, it is interesting to perform a comparative study between various categories. In the present paper we describe the comparison work performed between several statistical methods (LRA, ARIMA, PCA) and classification algorithms, as artificial neural networks, decision trees, conjunctive rules, support vector machines, decision tables and fuzzy lattice rules. Useful results are drawn concerning the performance and the operational potential of such methods.

Read also... Article (in pdf) - UAQ2005