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
  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)
#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:

#slightly cleaner looking summary table:
 post = my_post
 , par = 'some_parameter'


mike-lawrence/ezStan documentation built on June 20, 2018, 6:45 p.m.