Figure 3 | Scientific Reports

Figure 3

From: Disease severity classification using passively collected smartphone-based keystroke dynamics within multiple sclerosis

Figure 3

Example of feature selection using RFE schema with NHPT as a target. The x-axis shows the number of selected features, while the y-axis denotes the mean of AUC scores over k=5 folds for a given classifier. The final feature set is constructed by considering the most recurring features among all trained models where the mean AUC-ROC is maximized. One can also observe a performance plateau whenever more than eight features are used as predictors, regardless of the type of classifier.

Back to article page