This project is no longer being developed. See aria for a more reccent project with some shared/extended features.
An R package containing helper functions for Stan.
This package contains code I develop through my teaching and personal use of Stan/RStan to make some common tasks easier. It's very much a work in progress and probably won't ever be on CRAN.
Install via:
devtools::install_github('mike-lawrence/ezStan') #requires devtools to be installed, which is achieved via install.packages("devtools")
Optionally, for better warning & error message handling, install loggr:
devtools::install_github('mike-lawrence/loggr') #temporarily necessary until smbache/loggr is updated and uploaded to CRAN
Usage Example:
#compile a model; shouldn't block the RStudio UI like rstan::stan_model()
my_mod = build_stan(my_stan_file)
#start the model sampling
start_stan(
data = my_data_for_stan
, mod = my_mod
# can add other arguments to rstan::sampling() here if nessary, for example:
, iter = 4e3
)
#now watch the sampling (see ?watch_stan for options, like killing when post-warmup divergences are encountered)
watch_stan()
#it's ok to stop the watcher; you can resume watching by simply running watch_stan() again
#if anything goes awry with the sampling, you can call kill_stan() to stop all chains
#when the sampling is done, collect results
my_post = collect_stan()
#once collected, you can recover some disk space by running:
clean_stan()
#slightly cleaner looking summary table:
stan_summary(
post = my_post
, par = 'some_parameter'
)