Air quality prediction using optimal neural networks with stochastic variables

• Published in: Atmospheric environment (1994) Vol. 79; pp. 822 - 830
• Main Authors: Russo, Ana, Raischel, Frank, Lind, Pedro G.
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01-11-2013; Elsevier
• Subjects: Air quality; Analysis methods; Applied sciences; Atmospheric pollution; Climatology; Environmental research; Exact sciences and technology; Mathematical models; Neural networks; Optimization; Pollutants; Pollution; Stochastic systems; Stochasticity; Training; Weather; 02.50.Ga; Stochastic systems; Neural networks; Environmental research; 02.50.Ey; Pollutants; 92.60.Sz; Air quality; Neural network; Stochastic method; Statistical method; Atmospheric pollution forecasting; Air pollution
• ISSN: 1352-2310; 1873-2844
• DOI: 10.1016/j.atmosenv.2013.07.072

We apply recent methods in stochastic data analysis for discovering a set of few stochastic variables that represent the relevant information on a multivariate stochastic system, used as input for artificial neural network models for air quality forecast. We show that using these derived variables as input variables for training the neural networks it is possible to significantly reduce the amount of input variables necessary for the neural network model, without considerably changing the predictive power of the model. The reduced set of variables including these derived variables is therefore proposed as an optimal variable set for training neural network models in forecasting geophysical and weather properties. Finally, we briefly discuss other possible applications of such optimized neural network models.
•Optimized variables selection on a multivariate system by stochastic data analysis.•Selection of variables used as input for ANN air quality forecast models.•Use of derived variables as input for ANN models maintains forecast capabilities.•Use of derived variables as ANN's inputs reduces the amount of input variables.•Methodology can be adapted to other ANN models in weather or geophysical forecast.