Statistical Inference for Random Forests

Download .zip Download .tar.gz View on GitHub


This R package computes uncertainty for random forest predictions using a fast implementation of random forests in C++. Two variance estimates are provided: U-statistic based (Mentch & Hooker, 2016) and infinitesimal jackknife (Wager, Hastie, Efron, 2014).


Mentch L, Hooker G. Quantifying uncertainty in random forests via confidence intervals and hypothesis tests. Journal of Machine Learning Research. 2016.

Wager S, Hastie T, Efron B. Confidence intervals for random forests: the jackknife and the infinitesimal jackknife. Journal of Machine Learning Research. 2014.

Latest Version

v0.0.0.9000, under development


Package not yet on CRAN. For now, download from this website the .tar.gz (whether you are on Mac, Linux, or Windows). Do not download the .zip file. Note that both are source, not binary files. If you don't already have the dependencies (Rcpp, RcppArmadillo, Matrix, knitr) and optional dependencies (randomForest, rpart) installed, install those from CRAN first. If you already have an older version of surfin installed, remove that first by typing the following in R:

$ remove.packages("surfin")
Then install using:

$ install.packages(path_to_downloaded_file, repos=NULL, type="source")
$ library(surfin)
Learn how to use surfin from:

$ ?surfin

Authors and Contributors

Sarah Tan @shftan, David Miller @d-miller, Giles Hooker @gileshooker, Lucas Mentch


Questions, bug reports, etc.?