waternk/stan-book

Repository for book "Bayesian Statistics Using Stan", text files in Rmarkdown, plus example data, Stan models, and scripts

The Stan Book

This is the repository for *Bayesian Statistics Using Stan", which serves as both the Stan users' guide and an introduction to Bayesian statistics.

Directory Structure

• *.Rmd files: basic text
• _bookdown.yml: book includes
• _output.yml: output config
• bib/all.bib: BibTeX file for references
• stan/*.Stan : directory of Stan programs
• data/{*.R, *.rds} : directory for data used by programs
• programs/{*.R, *.stan} : legacy programs from old manual (deprecated until they're moved into new style with R inline in .Rmd)

Building the Book from Source

You will need to have RStan installed in the R environment from which you build.

RStudio

In RStudio: to build the project, open index.Rmd in RStudio and click knit - change output on first line of index.Rmd for gitbook and pdf_book (not differeing _)

Outside of RStudio

First, you will need to install pandoc and pandoc-citeproc in addition to the bookdown package in R. After that, it can be built from within R in this directory using bookdown::render('index.Rmd') or from the shell using ./build.sh to build both PDF and HTML versions.

Style Guide for Authors

• All lines should be 80 or fewer characters unless absolutely mandated by content

• y ~ normal(mu, sigma) # Not: N(), not sigma^2, regular font for "normal", Latex math for $y$, $\mu$, $\sigma$

• norma(y | mu, sigma) # Vertical bar, not semicolon

• Poisson, Weibull, LKJ # Use capital letters for distributions that are named after people

• E(y) # Regular font, parentheses not brackets

• () # Always parentheses, never brackets

• No special fonts for distributions, just latex roman and math fonts

• p(y) # Probability density and probability mass function

• Pr(A) # probability of an event

• Follow the Stan style guide for code

  • int<lower = 0> N; # Put in the lower bound
  • for (n in 1:N); # Not: for (i in 1:n);
  • foo_bar # Underscores rather than dots or CamelCase

• No R/Python code in the finished book except in appendix

• All Stan code should be best practice except when explaining something, in which case we should explicitly show the best-practice alternative

Licensing

The code is licensed under BSD-3 and the text under CC-BY ND 4.0.