README.md

In elizabethchase/HPR: Horseshoe Process Regression

HPR

This is the repository for the HPR R package, which accompanies Chase, Taylor, and Boonstra's paper on horseshoe process regression (HPR; 2022+). Interested users should examine that paper, along with the manual and vignette posted here, for examples and explanations for how to use the package.

Users should first install cmdstanr and cmdstan before attempting to install HPR. To do so, we first recommend starting a fresh R session. Then, run the commands install.packages("cmdstanr", repos = c("https://mc-stan.org/r-packages/", getOption("repos"))), followed by library(cmdstanr) and check_cmdstan_toolchain(fix = TRUE, quiet = TRUE).

If your toolchain is configured correctly, then run: install_cmdstan(cores = 2).

To check that cmdstan was installed correctly, run: cmdstan_path() and cmdstan_version().

At this point, you should be all set. If you have trouble with any of the steps above, please see the cmdstanr documentation for more information:

• https://mc-stan.org/cmdstanr/

• https://mc-stan.org/cmdstanr/articles/cmdstanr.html

After cmdstanr and cmdstan have successfully been installed, HPR can be installed by running the command install.packages("devtools") followed by devtools::install_github("elizabethchase/HPR"). At this point, you should be all set.

If you are still having trouble, we recommend checking that you have updated to the latest versions of both R and RStudio and then attempt reinstallation.

To get faster run-times from HPR, we recommend using parallelization by running the command options(mc.cores=3) prior to running any of the models in HPR. (3 cores is just an example; feel free to use more or less as your computer allows.)

elizabethchase/HPR documentation built on May 7, 2023, 5:48 a.m.