Creating reproducible reports with knitr and pandoc

These are some examples on how to use Markdown with R and pandoc to create dynamic documents for multiple output formats. All examples and accompanying text are contained in example.Rmd.

Compiling the document

Creating PDF and HTML output from the R/Markdown source file is a two step process. First knitr is used to execute the R code and produce the corresponding Markdown output. This can be done either by starting an R session and executing knitr("example.Rmd") or from the command line:

Rscript --slave -e "library(knitr);knit('example.Rmd')" 

Either way this generates a Markdown file called 'example.md'. This can then be converted into PDF and HTML files by using the configuration file 'example.pandoc' by calling the pandoc function from the knitr package.

Rscript --slave -e "library(knitr);pandoc('example.md')"

The function automatically locates the configuration file and passes the requested parameters to pandoc.

Required software

In addition to installations of knitr and pandoc a few external tools are required to compile this document.

R is required to run knitr as well as other R packages to support additional functionality.

Additional R packages used:

• animation (for animated figures)
• pander (for Markdown formatting of R objects)

These can be installed via the install.packages command from within R. Animations also require ffmpeg and either ImageMagick or GraphicsMagick.

As one might expect a working LaTeX tool chain is required to generate PDF output from LaTeX documents. Several distributions are available online, including MiKTeX and TeX Live.

Python ($\geq$ 2.7) is required for the pandoc filters discussed in the latter parts of this document. This also requires the pandocfilters Python module, which can be installed via pip.