Prediction of Missing Data for Ozone Concentrations using Support Vector Machines and Radial Basis Neural Networks

In this paper we present results from prediction of data for ozone (O3) concentrations in ambient air by using the modelling techniques of support vector machines (SVM) and radial basis neural networks (RBF NN). The predictions are performed for two short periods of time: for 24 hours and for one week in August and in December in 2005, in Skopje, Macedonia. The built SVM models use different kinds of kernels: polynomial and Gaussian kernels and the best values of the free parameters of SVM kernels are chosen by examining a range of values for each of the free parameters. This is the first attempt in Macedonia for prediction of concentrations of any air parameters in the ambient air.