Influence of settings and predictors in neural network model performance: a Buenos Aires air quality case

A. SCAGLIOTTI; D. MARGARIT; M. REALE; G.A. JORGE

Congreso; 11th International Young Scientist Conference on Computational Science; 2022

with crossed sensitivities. In the case of air quality, several antecedents seek to predict concentrations of pollutants, but generally,it is done with default Neural Network parameters and predictors selected with expert knowledge, which biases the results. Inregions with scarce air quality measurements, this problem is even more complex. This study aims to explore and present a novelmethodology for the design of a Multilayer Perceptron-type Neural Network for particulate matter prediction. Non-linear machinelearning hybrid methods are implemented for the selection of predictors using a testing bench Perceptron and Self-OrganizingMaps. The final model showed a better fit (correlation coefficient of 0.88 during the testing stage with new data and an root meansquared error of 1.8 μgm−3) than an expert model trained for the same study case and can be adapted for other regions and in otherfields of study.