Slides and blog posts with R and emacs org-mode

Preparing a larger number of slides with R code and plots can be a bit tedious with standard desktop presentation software like powerpoint or keynote. The manual effort to change the example code, run the analysis and then cut and paste updated graphs, tables and code is high. Sooner or later one is bound to create inconsistencies between code and expected results or even syntax errors

Using emacs and its swiss army knife org-mode there is another elegant and reproducible solution: just export the consistent code and output from an org-mode code block to generate either slides or bog pages (or both) from a single source. As a free benefit freen this tool chain can also produce high quality PDF handouts with a managed table of content and an index.

Org Mode source blocks

Here is a standard statistics “hello world” example using ggplot2:

1
2
3
4

  library(ggplot2)

  df <- data.frame( norm = rnorm(10000))
  ggplot(df, aes(x = norm)) + geom_histogram()

The source for the above blog output looks in org mode:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10

  ... some text describing the example...

  #+BEGIN_SRC R :exports both :results output graphics :file p.png
  library(ggplot2)

  df <- data.frame( norm = rnorm(10000))
  ggplot(df, aes(x = norm)) + geom_histogram()
  #+END_SRC

  ... more standard text...

Note the options in code block header that define whether source and output (including graphics) will be visible on the output page or slide. That’s all it takes to produce the post from the code and consistently re-create all output on the next publish step.

From blog entries to slides

To create visually appealing slides from org-mode there are several alternatives. For my not too complicated workflow I opted for reveal.js, the java script package by Hakim El Hattab. The installation and basic usage of this package from org-mode is described eg in this very nice tutorial by Mike Zamansky. With a few steps one can create a self consistent presentation file, which does not require any major software installations for the presentation viewer apart from a web browser.

To go from the above org-mode blog source to set of slides is only a small step:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11

  #+REVEAL_THEME: sky
  #+OPTIONS: toc:nil num:nil reveal_title_slide:nil
  * The Normal Distribution
  #+BEGIN_SRC R :exports both :results output graphics :file p.png :height 300
    library(ggplot2)

    df <- data.frame( norm = rnorm(10000))
    ggplot(df, aes(x = norm)) + geom_histogram()
  #+END_SRC
  - symmetric around $x = 0$
  - $P(x) = \frac{1}{{\sqrt {2\pi } }}e^{ - \frac{{x^2 }}{2}}$

And here is the result:

Just a png for now, as the blog publishing step needs a fix to allow accessing reveal.js from my blog hosting side.

• R
• Org
• Mode
• Emacs