Figure 1. Predictive performance of the machine learning using the Normal training datasets.

The letters F, C and F+C indicate that the Normal training datasets originated from the feature descriptors “frequency”, “composition” and “frequency” plus “composition”, respectively. The prime symbol indicates the Normal training datasets formed using the symmetrical attributes of the previously mentioned datasets (details in the “Material and Methods” section). The numbers at the bottom of the boxes are the medians for each dataset.