diversityForest: Innovative Complex Split Procedures in Random Forests Through Candidate Split Sampling

Implements interaction forests [1], which are specific diversity forests and the basic form of diversity forests that uses univariable, binary splitting [2]. Interaction forests (IFs) are ensembles of decision trees that model quantitative and qualitative interaction effects using bivariable splitting. IFs come with the Effect Importance Measure (EIM), which can be used to identify variable pairs that feature quantitative and qualitative interaction effects with high predictive relevance. IFs and EIM focus on well interpretable forms of interactions. The package also offers plot functions for visualising the estimated forms of interaction effects. Categorical, metric, and survival outcomes are supported. This is a fork of the R package 'ranger' (main author: Marvin N. Wright) that implements random forests using an efficient C++ implementation. References: [1] Hornung, R. & Boulesteix, A.-L. (2022) Interaction Forests: Identifying and exploiting interpretable quantitative and qualitative interaction effects. Computational Statistics & Data Analysis 171:107460, <doi:10.1016/j.csda.2022.107460>. [2] Hornung, R. (2022) Diversity forests: Using split sampling to enable innovative complex split procedures in random forests. SN Computer Science 3(2):1, <doi:10.1007/s42979-021-00920-1>.

Package details

AuthorRoman Hornung [aut, cre], Marvin N. Wright [ctb, cph]
MaintainerRoman Hornung <hornung@ibe.med.uni-muenchen.de>
Package repositoryView on CRAN
Installation Install the latest version of this package by entering the following in R:

Try the diversityForest package in your browser

Any scripts or data that you put into this service are public.

diversityForest documentation built on March 31, 2023, 9:58 p.m.