forestBalance: Balancing Confounder Distributions with Forest Energy Balancing

Estimates average treatment effects using kernel energy balancing with random forest similarity kernels. A multivariate random forest jointly models covariates, outcome, and treatment to build a similarity kernel between observations. This kernel is then used for energy balancing to create weights that control for confounding. The method is described in De and Huling (2025) <doi:10.48550/arXiv.2512.18069>.

Version:	0.1.0
Imports:	grf (≥ 2.3.0), MASS, Matrix, methods, Rcpp
LinkingTo:	Rcpp, RcppEigen
Suggests:	ggplot2, knitr, osqp, rmarkdown, testthat (≥ 3.0.0), WeightIt
Published:	2026-04-07
DOI:	10.32614/CRAN.package.forestBalance (may not be active yet)
Author:	Jared Huling [aut, cre], Simion De [aut]
Maintainer:	Jared Huling <jaredhuling at gmail.com>
BugReports:	https://github.com/jaredhuling/forestBalance/issues
License:	GPL (≥ 3)
URL:	https://github.com/jaredhuling/forestBalance
NeedsCompilation:	yes
CRAN checks:	forestBalance results

Documentation:

Reference manual:	forestBalance.html , forestBalance.pdf
Vignettes:	Augmented (Doubly-Robust) Estimation (source) Cross-Fitting for Debiased Kernel Estimation (source) Getting Started with forestBalance (source) Performance and Scalability (source)

Downloads:

Package source:	forestBalance_0.1.0.tar.gz
Windows binaries:	r-devel: not available, r-release: not available, r-oldrel: not available
macOS binaries:	r-release (arm64): not available, r-oldrel (arm64): not available, r-release (x86_64): not available, r-oldrel (x86_64): not available

Linking:

Please use the canonical form https://CRAN.R-project.org/package=forestBalance to link to this page.