The Classification of Valid and Invalid Beats of Three-Dimensional Nystagmus Eye Movement Signals Using Machine Learning Methods
Figure 8
The two greatest principal components were used to map how accepted and rejected nystagmus candidates by manual selection were distributed before (a) and after (b) cleaning (situations (1) and (2) from Figure 6). No more than one tenth of nystagmus candidates are mapped to make the figures clear, since the rejected and accepted candidates overlap and the rejected candidates are strongly concentrated in the “left corner” of the distribution. The two principal components corresponded to (a) 80.0% and (b) 82.1% from variance in the data.
