Results of the Gaussian mixture modeling (GMM) given as , where mi, si and wi are the parameters mean, standard deviation and relative weight of each of the Gaussians, respectively, obtained for the UGini index data.

The final model with an optimum number of M = 4 mixes was selected on the basis of the Akaike information criterion [26], visual inspection of the fit, the statistical significance of an χ2 test estimating the likelihood that the final model did not adequately describe the data, and a quantile-quantile plot (Fig 5).