Encouraging results were achieved, with the SVM model providing the best
performances, outperforming the NN and MR techniques. The overall accuracies
are 64.3% (T = 0.5) and 86.8% (T = 1.0). It should be noted that the
datasets contain six/seven classes (from 3 to 8/9) and these accuracies are much
better than the ones expected by a random classifier. While requiring more computation,
the SVM fitting can still be achieved within a reasonable time with
current processors. For example, one run of the 5-fold cross-validation testing
takes around 26 minutes.