For the two prediction models, OGP and FOM, the following assumptions were checked, by means of specific statistical tests: the multicollinearity was evaluated by examining the Pearson product–moment correlation factor; the linearity of the relationship between dependent and independent variables was measured by correlation coefficients and examination of the partial regression plots; the independence of the errors was checked using the Durbin-Watson statistic; the homoscedasticity of the errors was tested through a Lagrange Multiplier test; and the normality of the error distribution was appraised