Skip to main content
. 2013 Sep 18;25(5):1804–1823. doi: 10.1177/0962280213502437

Figure 6.

Figure 6.

Leave-one-out CV errors for classifying 86 patients into ADHD and Control groups based on top EEG measures. This is an example where the effect size is very small and therefore the error rate does not show a steady declining pattern as the number of variables grows. This however supports the observation from simulation regarding the strengths of RF in dealing with data having weaker signal and higher noise as RF is seen to outperform LDA and kNN in this example.