Stochastic tree ensembles (BART / XBART) for supervised learning and causal inference.
To install any of the options below, you must first clone the repository to your local machine. To do that, you must have git installed, which you can do following these instructions.
Once git is available at the command line, navigate to the folder that will store this project
(in bash / zsh, this is done by running cd followed by the path to the directory).
Then, clone the StochasticTree repo as a subfolder by running
git clone --recursive https://github.com/andrewherren/StochasticTree.git
NOTE: this project incorporates several dependencies as git submodules,
which is why the --recursive flag is necessary (some systems may perform a recursive clone without this flag, but
--recursive ensures this behavior on all platforms).
The R package is defined in the R-package subfolder of StochasticTree but it depends on
C++ code in the main project folder (as specified in R-package/src/Makevars).
There are several ways to install the R package after cloning the repo locally.
From the R console, you can nagivate to the main project directory
(i.e. setwd("/path/to/StochasticTree")) and then run
install.packages(pkgs = "R-package", repos = NULL, type = "source")
From the command line, navigate to the main project directory (i.e. cd /path/to/StochasticTree)
and then run
R CMD INSTALL --preclean R-package
The python package can be installed from source. Before you begin, make sure you have conda installed. Clone the repo following the instructions in the "cloning the repository" section above.
Next, create and activate a conda environment with the requisite dependencies
conda create -n stochtree-dev -c conda-forge python=3.10 numpy scipy pytest pandas pybind11
conda activate stochtree-dev
conda install -c conda-forge matplotlib seaborn
pip install jupyterlab
Then, navigate to the main StochasticTree project folder (i.e. cd /path/to/StochasticTree) and install the package locally via pip
pip install ./python-package
The command line interface can be built from source using cmake.
See here for details on installing cmake (alternatively,
on MacOS, cmake can be installed using homebrew).
Once cmake is installed, you can build the CLI by navigating to the main
project directory at your command line (i.e. cd /path/to/StochasticTree) and
running the following code
rm -rf build
mkdir build
cmake -S . -B build
cmake --build build
One goal of this project is to provide a unit-tested, low-level core of C++ data structures needed to build stochastic tree models, so that researchers / implementers don't have to "reinvent the wheel." As we develop and stabilize a C++ interface, we will document our recommendations for using it in your own C++ project.
The R-package subfolder includes two demo scripts (demo/xbart_demo.R and demo/bart_demo.R) which can be run at the R console in RStudio.
The python-package subfolder includes two demo notebooks (demo/xbart_demo.ipynb and demo/bart_demo.ipynb) which can be run via your preferred jupyter environment (browser, VS code, etc...).
Once built, the CLI can be run directly from the command line,
with configuration options specified in a .conf file.
See demo/xbart_train/ and demo/bart_train/ for examples.