ESTUDO DA QUALIDADE DO AR DO ESTADO DE S O PAULO POR MEIO DE MAPAS AUTO-ORGANIZáVEIS DE KOHONEN

The present study uses Kohonen self-organizing maps of to analyze the air quality in the stateof S o Paulo. The data source was the CETESB - Environmental Company of S o Paulo. Weuse the Kohonen neural network to group the cities into groups that have homogeneous levelsof pollution, i.e. that have a similar profile of distribution of pollutants in the atmosphere. Theuse of Kohonen neural network helps determine a minimal architecture, capable ofrepresenting the data set. The article identified groups of municipalities which, although notphysically close, belong to the same group of pollution.