The Quality of Scientific Plotting Software

I’m going to take a minute here to rant a little bit about the paltry menu of scientific plotting software that is available.

This research project that I’m working on has required me to teach myself the basics of multivariate linear regressions and a few other things. I’ve finally gotten something that I think is legitimate, although I won’t know until tomorrow, when I meet with the physician I’ve been working with. I want to make some simple box-and-whisker plots that show the differences between a few different groups. I’ve been using JMP to do my analysis, but I’d like to produce some publication quality figures so that I can start working on our paper. The problem is that the options for generating publication quality figures absolutely sucks. I’ve spent the past six hours trying to figure out the best way to generate a half dozen plots that don’t look like they were cobbled together in Excel and I’ve gotten nowhere. Screw all of this.

This isn’t the first time I’ve had this type of frustration. Here are the options I’ve tried, that produce crappy results.

  • Matlab is nice for working with large datasets, particularly if you’re doing things like Fourier analysis or numerical modeling.  The problem that Matlab has is that it can’t produce reliable figures – what you see on your screen is usually not what you get when you generate an EPS or TIF.  Also, what the hell is with medical journals that don’t want vector images?  I was blown away when I found out that the journal I’m hoping to submit wanted GIF or JPG figures.  At any rate, Matlab is terrible.
  • JMP is fine for producing quick plots of various statistical relationships, but its ‘Graph Builder’ is absolutely useless.  Beyond useless.  Honestly, the vast majority of statistical software I’ve used has the most worthless user interface.  It’s enough to make me think that GUI statistical software can shove it.  If I get into a MD / PhD program, I swear I will never use any of this crap – I’ll teach myself to use R and never look back.
  • SigmaPlot.  This piece of crap costs a zillion dollars and the interface is nearly impossible to use.  It’s as if these packages are designed by people that don’t actually know anything about how data analysis is done.
  • Microsoft Excel.  I hate almost everything Microsoft does, but Excel 2003 was awesome.  Of course, after that, Microsoft changed the interface entirely that is completely useless and makes the simplest tasks a Herculean epic.  Of course, there’s the problem of no capacity to export into a usable format.  Worthless

Seriously.  This is infuriating.  It should not be this difficult to produce properly formatted and sized figures.  Even the most expensive programs do an abysmal job.  All I want to do is make a few box-and-whisker plots with well-defined resolutions, fonts, formats, and sizes.  I have wasted countless hours trying to do this and I’m absolutely fed up with it.  Seriously, fuck this.

About these ads

11 Responses

  1. Have you tried SPSS (a SAS product)? It produces some nice graphs. It’s not as easy to use as MATLAB, but I like it.

    I used it when I took Statistics and have used it in the lab I work in on campus to analyze data from my project. The graphs I created will be used in publication, so might be good enough for you to use.

    Let me know if you want more info!!!

    • Yeah, I thought about SPSS too. It’s hugely expensive though. I think I’m just going to teach myself to use R. It can generate great looking figures from what I’ve seen and seems to be what most statisticians prefer anyway.

  2. try using minitab? i think we use that for various six sigma stuff…

    • Thanks for the suggestion. I decided to just teach myself R. Stats people all over the planet use R, so I figure it makes sense to learn it now, rather than later.

      The syntax is a little weird, but it doesn’t seem too bad. So far, anyway.

  3. Hi,

    We develop a 2D/3D data plotting and curve fitting program called TeraPlot:

    http://www.teraplot.com.

    We like to think the software has the right combination of functionality and ease of use. If you were minded to investigate it, I’d be very interested in any comments you might have.

  4. Of all the scientific graphics and statistics software I have tried,GraphPad Prism 6 offers the best combination of features and a highly intuitive interface. It is available for both Mac and Windows. The next best program is OriginPro 9 — it has more advanced features than GraphPad Prism, but it is not as intuitive and runs only on Windows.

  5. I use Magic Plot both on Linux and Mac, it works pretty nice!

    http://magicplot.com

    • I’ve concluded that about the only game in town is something like Matplotlib, R, or Matlab, if you’re willing to spend the time to make it work. Why anyone performs data analysis on Windows is beyond me.

  6. If you’re familiar with Matlab, ILNumerics might be interesting for you, too. It’s a math library for .NET that provides a very convenient syntax and nice plotting features. You can try ILNumerics online: http://ilnumerics.net/ilnumerics-interactive-web-component.html

  7. Have u tried dplot.com?
    $200, box & whisker plots

  8. I’m a big fan of Veusz. It is available on Linux, Mac, and Windows. It is open source and uses a GUI.

    http://home.gna.org/veusz/

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 60 other followers

%d bloggers like this: