How reliable is the Environmental Sustainability Index 2005?

The Environmental Sustainability Index (ESI) 2005 is investigated in terms of its reliability, namely whether (and how well) the ESI score of a country of interest can be evaluated provided that the necessary data are available. A general regression artificial neural network (GRNN) architecture is used to this end, selected on the grounds of its swift training and ability to non-parametrically follow any given (hyper-)curve based exclusively on the available data. The results obtained from two cross-validation schemes (leave-one-out and ten-fold) which have been applied to the 146 countries that participated in the creation of the ESI 2005 indicate that the accurate prediction of the ESI score of a novel country (with data comparable to those of the participating countries) and, consequently, its ranking relative to the participating countries cannot be guaranteed.