Why aren't all of our graphs interactive?

During the last pre-happy hour seminar, Karl Broman talked about Why aren’t all of our graphs interactive? I didn’t know, but a few years ago Karl worked in the department and clearly promoted beer-drinking and is the heart of the department. I’m a fan of our pre-happy hour seminars since you have a get to listen to good/fun talks over a beer or two.

But I’m also a fan of reproducible research and useful graphics. I do most of this by using Sweave (for reproducibility) in LaTeX documents and with the R packages lattice, car, and plotrix, and some ggplot2 (I should use it more). 

Karl made his presentation using html (definitely check it out!) and inserted pretty interactive graphics. His talk got me really interested and I definitely need to pick up a few tools. For example, asciidoc or R Markdown can be useful for making html documents with R code. Specially if you want to write a report and you don’t want to deal with Sweave/Latex when making plots (can be a pain to know where they’ll show up). 

For the interactive side, D3 (and other tools Karl listed) can be useful to learn. But I might put this on a hold for some time. Maybe I’ll wait and see what others in the deparment are developing for R-D3 and embedding interactive plots in pdf files.

I don’t think that it will be long before interactive plots make it to the journals. Specially for their web versions. Though, I still think that if you are showing a 3D plot, as the author you will have to give a few default views where you can clearly see something that you want to talk about instead of having the reader find that sweet spot. 

One problem that I don’t think has been solved yet is reproducible research on a cluster. Karl and others mentioned make as well as having if/else clauses where you either show the output or a cleaned up version of the code that you used to generate the output. 

Overall, there are many tools and tips I can learn from Karl. And I’m sure that I’m not the only one! Hopefully he’ll give tips on where to start (nothing is more tedious than reading UNIX man-files).

comments powered by Disqus