Fig. 4From: Automatic sleep staging using ear-EEGCohens Kappa (\(\kappa\)) for different numbers of stages and different ways to cross validate. For comparison, the percentage of correctly labeled stages is also shown. Not surprisingly, we see that \(\kappa\) increases when the number of stages decrease, and when the classifier has more prior information about the subject(s). In all cases, for calculation of both Kappa and accuracy, all epochs have been pooled together before calculation, equivalent to taking a population average weighted by number of subject epochsBack to article page