Table 1.

Accuracy of principal features subspace method for different splits of training and test samples.

Train/test split	Train accuracy (%, mean ± std)	Test accuracy (%, mean ± std)
80/20	96.23 ± 2.24	93.11 ± 3.61
50/50	96.30 ± 2.59	92.94 ± 3.82
30/70	96.81 ± 3.07	90.23 ± 4.30
20/80	97.01 ± 3.22	87.60 ± 5.27
10/90	97.72 ± 2.65	81.86 ± 7.15

The results show consistently high train accuracy across a range of training set sizes. As expected, training accuracy increases with smaller training sets; however, test accuracy goes down as training set size is reduced, due to overfitting.