Figure 6

Structure of the CLASSIFIER node. The CLASSIFIER node has attributes for naming the classifier, indicating the classification method and the creation date; and it is composed by a sequence of six nodes: Dataset, Classes, Boundaries, Features, Weights, and EvaluationResults. The Dataset node has only the path to the dataset file. The Classes node contains a series of Class nodes for storing the tumour types involved in each class. The Boundaries node is for storing the points that form the boundaries between classes in the projection space: they are the intersection point (IntersectionPoint node) and the rest of points (the Point node sequence) used to draw a line from each of them to the intersection point. The Features node has the attribute Method for the name of the FS/FE method used, and the list of the resulting features. The Weights node contains the sequence of weights of the classifier, and the associated feature to each of them. The EvaluationResults node is for storing information related with the evaluation of the model, in this case, using bootstrapping (the Bootstrapping node) and the ROC curve (the AUC node). The Bootstrapping node has two attributes for the overall mean and standard deviation, and a list of nodes with the bootstrapping results per class. The AUC node contains a sequence of nodes with the AUC results by class. Dashed lines are used to indicate non-mandatory elements.