When comparing the classification matrices of both NN, mine and SSAS, the global error was quite similar (mine was a little higher but not much). The intention is to show these results when advancing in a coming article

Thanks for the comments!

Silvia