Resumo

MUSSO, Haydeé Elena  e  AVILA BLAS, Orlando José. The  use of multilayer  perceptrons  for statistical  modeling so₂  non linear time  series in Salta  Capital,  Argentina. Rev. Mat [online]. 2013, vol.20, n.1, pp.61-78. ISSN 1409-2433.

In  this paper a statistical study of phisical-chemistry variables connected with enviroment pollution, specifically SO₂ monthly average concentration, measured in Salta Capital city, Argentina, together  with  NO₂  and  O₃  concentrations,  was made.  Time  series under study shown non linear dinamic behaviour, outliers and structural changes. Due to these it was impossible to use typical econometric  typologies (AR,  MA, ARMA,  ARIMA,  among  others).   An effective solution which uses multistep perceptrons theory was found. By using structural time series modelling, this solution is presented by an iterative mathematical process that  allows us to obtain  a final  model  with  a  high  confidence  level  (95%)  in  order  to do  the forecasting step on the studied variable.

Palavras-chave : time series; modelling; multistep perceptrons; air pollution; sulfure dioxide; passive sampling.

        · resumo em Espanhol     · texto em Espanhol     · Espanhol ( pdf )