The multi-way importance plot (Figure 1) displays all of the significant variables (labeled and marked with black circles) that contributed to the classification of the memory group from the control subjects; they are ranked based on their importance for classification as derived from Gini decrease, number of trees, and p-value. A chart shows the distribution of minimal depth in classification against the number of decision trees (see Figure S4 in the Supplementary Materials). While both a multi-way importance plot and a distribution plot can be created for any set of random forest parameters, the importance ranking for the features is likely to be similar owing to high correlations among these parameters (see Figure S5 in the Supplementary Materials).
