grf: Generalized Random Forests

Forest-based statistical estimation and inference. GRF provides non-parametric methods for heterogeneous treatment effects estimation (optionally using right-censored outcomes, multiple treatment arms or outcomes, or instrumental variables), as well as least-squares regression, quantile regression, and survival regression, all with support for missing covariates.

Version:	2.6.1
Depends:	R (≥ 3.5.0)
Imports:	DiceKriging, lmtest, Matrix, methods, Rcpp (≥ 0.12.15), sandwich (≥ 2.4-0)
LinkingTo:	Rcpp, RcppEigen
Suggests:	DiagrammeR, MASS, policytree, rdrobust, survival (≥ 3.2-8), testthat (≥ 3.0.4), knitr, rmarkdown
Published:	2026-03-04
DOI:	10.32614/CRAN.package.grf
Author:	Julie Tibshirani [aut], Susan Athey [aut], Rina Friedberg [ctb], Vitor Hadad [ctb], David Hirshberg [ctb], Luke Miner [ctb], Erik Sverdrup [aut, cre], Stefan Wager [aut], Marvin Wright [ctb]
Maintainer:	Erik Sverdrup <erik.sverdrup at monash.edu>
BugReports:	https://github.com/grf-labs/grf/issues
License:	GPL-3
URL:	https://github.com/grf-labs/grf
NeedsCompilation:	yes
SystemRequirements:	C++17, GNU make
In views:	CausalInference, Econometrics, MachineLearning, MissingData
CRAN checks:	grf results

Documentation:

Reference manual:	grf.html , grf.pdf
Vignettes:	An introduction to grf (source, R code)

Downloads:

Package source:	grf_2.6.1.tar.gz
Windows binaries:	r-devel: grf_2.6.1.zip, r-release: grf_2.6.1.zip, r-oldrel: grf_2.6.1.zip
macOS binaries:	r-release (arm64): grf_2.6.1.tgz, r-oldrel (arm64): grf_2.6.1.tgz, r-release (x86_64): grf_2.6.1.tgz, r-oldrel (x86_64): grf_2.6.1.tgz
Old sources:	grf archive

Reverse dependencies:

Reverse imports:	aggTrees, causalDT, causalQual, causalweight, cohetsurr, cramR, EpiForsk, evalITR, HCPclust, htetree, longsurr, OutcomeWeights, policytree, qeML, roseRF
Reverse suggests:	CRE, evalHTE, maq, rdss, targeted

Linking:

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