Michael R. Head (suppressingfire) wrote,
Michael R. Head

Gnuplot vs. GNU R

I'm getting ready to do some data analysis for a paper (after I collect some performance numbers tomorrow), and I'm considering how I want to plot these things. In the past, I've relied almost exclusively on a combination of python and GNUplot. This has been convenient because the python makes it easy to parse data output files and generate GNUplot data and script files and it's been pretty easy to tie it all into a system of Makefiles so, e.g., "make graph" can compile, build, test, benchmark, analyze, and generate tons of PS and PDF graphs for the papers. Not bad, but I've reached the limits of these techniques and I'm interested in learning something new for my current project.

After browsing a technical report (that made its way across Slashdot) from Google labs detailing how SMART sucks, I noticed that they used GNU R (www.r-project.org) to analyze their data. I've seen the packages in Debian and Ubuntu for some time, but never had any desire to look at it, perhaps because of a certain snootiness about statistics in general (after all, it's just math for the "lower" sciences, right? hehe). Anyway, it looks like it can draw pretty graphs and if R scipts can eliminate my current python+gnuplot amalgam and create graphs better suited for publication, I'm all about picking it up. Plus, it'll help me learn a new technology on this project, which makes it a lot easier to stay focused.

So, is R worth it? Does it make the task of parsing, analyzing, and graphing data approach triviality? I guess I'm mostly looking in metahacker 's direction, since he's got the most statistics-fu of the folks I know...
Tags: analysis, statistics

  • Post a new comment


    default userpic

    Your reply will be screened

    Your IP address will be recorded 

    When you submit the form an invisible reCAPTCHA check will be performed.
    You must follow the Privacy Policy and Google Terms of use.