Being able to press a single button that runs all your statistical analyses and integrates the output into your final report is a beautiful thing. If you have not already heard, this is what Sweave can do for you. However, getting your computer to run Sweave can be a little bit fiddly. Thus, this post: (1) sets out the benefits of Sweave; (2) sets out how to install and configure R, Sweave, and Eclipse on Windows; (3) lists resources for people wanting to learn more about how to use LaTeX and Sweave; and (4) lists some specific resources relevant to researchers in psychology wanting to use these tools.
What is Sweave?
To Sweave is to weave in S. To weave is to combine data analysis code and standard formatted text into a single self-describing document. R is a dialect of S. Thus, if you use R to do your statistical analyses and you want to automate the importation of analyses in R into your reports, Sweave may be the tool for you. For a longer description, see Friedrich Leisch’s (2002) Sweave: Dynamic Generation of Statistical Reports Using Literate Data Analysis.
- Reproducibility: The most important reason to adopt a tool like Sweave is to make your research more reproducible. The R code sets out exactly how the raw data is transformed into publication output. The Sweave document links this R output with the final report.
- Efficiency: Statistical output is automatically incorporated into your report. There is no need to copy and paste output from your statistical analysis program into your report. If your data or analyses change, you can update your report with a single click instead of having to manually update every table and figure.
- Reliability: The integration of analyses with the report reduces the chance of errors entering in through copying and pasting of statistical output into documents.
- Education & Communication: By providing data analysis code for a report, this teaches others how to do similar analyses.
- For an extended discussion, see Anthony Rossini and Friedrich Leisch’s (2003) working paper Literate Statistical Practice.
Common Use Cases
- Statistics Instructional Materials
- Empirical reports, journal articles, book chapters, theses, etc.
Data sharing, literate programming, reproducible research, weaving: This is future of data analysis. Why not get on board now?
MY SWEAVE INSTALLATION AND CONFIGURATION
A strength of R, Sweave, and LaTeX is that they are cross platform tools that can be integrated together to support powerful data analytic workflows. These tools run on Windows, Linux, and Mac OS with a range of text editors and command line options. However, the flexibility in configurations presents a challenge. There is no single click installation file like “setup.exe”. The tools need to be assembled. This section sets out how to install and configure a system for writing Sweave documents based around the Eclipse IDE and a Windows Operating System. It’s not the only way to assemble a system to support Sweave, but for someone entrenched in the Windows world, I think its a good start.
1. Download and install R
2. Download and install a Latex distribution
There are several LaTeX distributions. I installed MikTex.
3.a Download and install Eclipse and the StatET and TeXlipse plugins
See the StatET Installation page for instructions on how to install StatET and Eclipse.
See also the links under “Getting Started” with StatET and Eclipse.
3.a Download and install Eclipse
See StatET Installation page; This assumes you have Java installed.
3.b Install the StatET plugin and the TeXlipse plugin
3.c Configure the StatET plugin
3.d Configure the TeXlipse plugin
The TeXclipse homepage lists general information. See specifically, the configuration page. My configuration could be abbreviated to: Go to Window – Preferences in Eclipse; Then, TeXlipse – Builder Settings; Then, enter the appropriate directory for your Bin directory of TeX distribution. In my case this was “D:\MiKTeX 2.8\miktex\bin” .
3.e Configure Sweave
- Sweave.sty: Sweave is a built-in function in R. However, when you run Sweave, your LaTeX distribution needs to be able to find a file called “Sweave.sty”. The file is stored in your R program files (e.g., “C:\Program Files\R\R-2.9.1\share\texmf”). A quick way to make it accessible is to place the file in your Eclipse project folder where the Rnw file is located. See this R-Help post for tips. UMN has some additional tips. (UPDATE: Bernd referred me to some additional material on linking Sweave.sty with MikTeX.
- External Tools: Go to Run — External Tools – External Tools configurations;
Sweave Document Processing (R/LaTeX); Click New Button; Give it a name like “Sweave-PDF”
Under the LaTeX tab change output format to pdf build commands pdflatex.exe
Using Sweave assumes that you know how to use LaTeX. If you just want to write LaTeX documents using Eclipse (without R code), you can go to File – New Project (Texlipse – LaTeX Project). Once you have a basic working environment, it’s easy to experiment with all the details of LaTeX. Here are some web guides among the many that are available.
Sweave is fairly straightforward. In Eclipse you can start a new R Project and add an *.Rnw file to write your Sweave document. Then use the Document menu to convert the Sweave file to a TEX file, PDF file, etc. There are many more general resources on Sweave:
- Friedrich Leisch’s Resources Page: This includes an overview of Sweave, Sweave documentation; example Sweave documents, and discussion of literate programming
- Learning to Sweave in APA Style: A useful introduction for those new to LaTeX, Sweave, and R.
- Nicola Sartori’s Sweave demo: 30 slide presentation introducing Sweave.
- Sarah Haile’s Overview: 17 slide introduction to Sweave.
- Charles J. Geyer’s Demonstration: Tutorial and notes with examples on using Sweave.
- Vinh Q. Nguyen’s UCI Seminar on Sweave: Slides introducing Sweave: Note particularly, the discussion of the
cacheSweavepackage for computationally intensive code that you don’t want to re-run every time you refresh your document.
- Benjamin Bolker’s website for his book Ecological Models and Data in R. The site has earlier drafts from his book and problem exercises in pdf and Sweave-ready format (see the Rnw files).
- Drew Conway referred me to Michael Malecki‘s Sweave Template
LATEX, SWEAVE, AND PSYCHOLOGY
Adopting LaTeX and Sweave presents several challenges related to somewhat discipline specific needs. These pertain particularly to the various style conventions expected for journal submission. The following are some useful resources:
- Introduction: Learning to Sweave in APA Style is A useful introduction if you are new to LaTeX, Sweave, and R from the perspective of a psychology researcher.
- APA Style: William Revelle provides a page on Writing a research methods paper in APA style using LaTeX
- APA Style assumes decimal point alignment in tables: dcolumn may be useful.
- Reference Manager for latex: bibtex is the tool; apacite provides APA style instructions to bibtex.
- Slide Creation: beamer is a commonly used tool to produce slides using LaTeX
- Spell checker for LaTeX in Eclipse: Martin Homik provides some suggestions.
- R and Psychology: My post on Getting Started with R for researchers in Psychology
- Options for Collaborating with Non Latex Users:
- Export Sweave document to HTML (e.g., R2HTML).
- Use odfWeave. It’s like Sweave, but it’s for Open Office.
- Convert LaTeX generated pdf to HTML.