3D Scatterspace Page

This page describes a project I've been working on, using Virtual Reality Modeling Language (VRML) version 2.0 to create 3D scatterplots (which I call "scatterspaces"). The advantage of these scatterspaces is that they can be manipulated and navigated in ways that traditional 3D spinplots can't--all using off-the-shelf web browser technology.

Examples of things you can do with the scatterspace:

How it works

From SPSS 7.5 for Windows, you run a script that I wrote (which should be available for download soon--but not yet). The script allows you to select variables from your active data file, as well as options (for rescaling variables and plotting the regression plane), all with a standard dialog box interface. Then the script goes and gets the data from the data file, makes whatever calculations are necessary, and generates the VRML for the scatterspace. Then it automatically calls up the appropriate browser (assuming your system is configured to recognize the .WRL extension as VRML 2.0) to render the scatterspace.

What you need

To generate a scatterspace:

To generate these using my script, of course you'll need SPSS 7.5 for Windows. You can download the script at http://www.execpc.com/~helberg/ScatterSpace.sbs. (If the script file comes across as plain text, just save it with the .SBS extension, and it should work fine.) Then, just run the script, and you've got your scatterspace!

To view a scatterspace:

You'll need a VRML 2.0 compliant browser or plug-in to view these scatterspaces. Most VRML 2.0 browsers are still under development. The best viewer I've found for this particular application is SGI's Cosmo Player.

A sample

Here is a sample, which I generated directly from the script. It is based on a subset of countries from the WORLD95.SAV dataset distributed as an example file with SPSS. The model here is urbanization (URBAN) and literacy rate (LITERACY) as they predict infant mortality (BABYMORT). Note that each data point is labeled for the country it represents, and it is color coded by climate (red=cold, orange=temperate, green=tropical, blue=missing). The regression plane reveals that there is a strong relationship between LITERACY and BABYMORT, and a weaker relationship between URBAN and BABYMORT.

Questions or comments

If you have questions or comments about this application of VRML, please send them to me at chelberg@spss.com. Your feedback is appreciated.