Skip to main content
Fig. 1 | BMC Bioinformatics

Fig. 1

From: Do little interactions get lost in dark random forests?

Fig. 1

Problematic splits for classification trees and random forests. In (a) no reasonable first split on the variables x 1 or x 2 can be made. However, two subsequent splits on x 1 and x 2 split the data perfectly. In (b), again no reasonable first split is possible, even though the classes are linear separable. Both the variables x 1 or x 2 have to be considered simultaneously and even with several subsequent splits on x 1 and x 2, no accurate classification is possible

Back to article page