Skip to main content
Fig. 4 | BMC Bioinformatics

Fig. 4

From: Visually guided classification trees for analyzing chronic patients

Fig. 4

Combination of SC and LDA for CRGs 5192-6144-7071. The mapped points (dots) correspond to an LDA plot, which separates the classes in an optimal way. In addition, the SC radial axis vectors (the specific set of vectors is chosen in order to generate the LDA mapping when applying SC) provide information about the features. Their length and orientation indicate the contribution of the features to the LDA class separation task. In this case the longest vector corresponds to the diagnosis code ‘250’, and points to the direction that better separates the points in CRG-5192 (orange) from the rest. The plot also shows that points of the CRG-6144 (green) and CRG-7071 (blue) classes are highly overlapped

Back to article page