Open Source Data Analysis – Oldies But Goodies
Today’s NYT has an article in its Business Computing section on R the open source data analysis language that’s starting to put a dent in SAS. Read the SAS spokeperson’s comments – looks like a case of open source denial in full bloom.
Secondly, a new book on Gnuplot is available for pre-release in PDF from Manning – Gnuplot in Action. I would highly recommend the book even though it’s not in final form and you have to buy the PDF from Manning directly. It’s basically complete and very usable – tons of good examples in it – and I first learned of R from this book’s mention of it.
I used Gnuplot extensively recently when I built test software for a wifi RF signal strength vs distance/terrain field study I did for a major construction equipment manufacturer. After getting into it, I wanted to write a blog entry called “Oldies but Goodies” and Gnuplot was going to be on the list.
Gnuplot has been around for a long time, and considering its maturity has continued to keep boundaries on itself so it’s not trying to be all things to all people. It’s not trying to be an R, for example. It’s a mark of a good OSS project when it can maintain that type of discipline.
FYI – here’s a quick glimpse of a Gnuplot that came from my wifi field study. Nothing fancy, but just nice, clean charts.
The beauty of using Gnuplot is how scriptable it is and how many output formats it can support…everything from ASCII art, to all the major bitmap graphic formats, to SVG, PS and PDF. I used Ruby to drive Gnuplot results for dozens of charts of different RF experiments done for that study. All the charts where generated by Gnuplot into PDF format so I could scale them easily. They could then be sucked into Mac Pages to build a very professional looking report. As PDFs, every chart was scalable once in the page layout program, so it was a perfect output format and could be quickly generated. As data changed during the experiments, I regenerated dozens of resulting charts effortlessly.
Gnuplot was a wonderful tool for RF data analysis and a reminder that not everything that’s great has to be hot and new.
[update: found this presentation on using Gnuplot and R together]
About me
I’m an independent consultant who used to do a lot of work for OpenLogic. I greatly appreciate OpenLogic and the mission they are fulfilling as well as their willingness to let me contribute to this blog. Views expressed here are not necessarily those of OpenLogic and any mistakes are 100% attributable to me. You can contact me at: landon at 360vl dot com or visit http://sawdust.see-do.org


