C443: An R-package to see a forest for the trees
Aniek Sies, Iven Van Mechelen · The R Journal · 2023
Classification trees, well-known for their ease of interpretation, are a widely used tool to solve statistical learning problems. However, researchers often end up with a forest rather than an individual classification tree, which implies a major cost due to the loss of the transparency of individual trees. Therefore, an important challenge is to enjoy the benefits of forests without paying this cost. In this paper, we propose the R package C443. The C443 methodology simplifies a forest into one or a few condensed summary trees, to gain insight into its central tendency and heterogeneity. This is done by clustering the trees in the forest based on similarities between them, and on post-processing the clustering output. We will elaborate upon the implementation of the methodology in the package, and will illustrate its use with three examples.