The month of May brought several remarkable software releases from two of my favorite companies, as well as a first experience with a new version of Windows from everyone’s favorite whipping boy, Microsoft. A full review of Wolfram’s new Mathematica 6 appears in this issue, and a review of SAS’ new JMP 7 is coming soon. For now, I’d like to share a little enthusiasm for both new versions that are far from the simple cosmetics that go into the N.x to N.x+1 upgrades.
Let’s start with JMP. This is statistical software that can be profitably and easily used by non-statisticians. Your editor was privileged to hear an introduction to version 7 by none other that John Sall, the original developer of the software and Executive VP of SAS. As this was at a local user’s group meeting, other speakers from JMP presented special topics using the new version so the audience was able to gain insight as to the power and flexibility of this new system, as well as to gage the company’s ability to examine and understand its customers’ needs.
JMP has long been known for its dynamic linking of instant graphics to analysis, and JMP 7 not only extends the variety of graphics, but the power as well. The new motion-enabled feature provides animation whereby variable changes may be instantly visualized and allows, for example, rapid recognition in patterns over time. A bubble plot that is multivariately-colored exhibited subset movement that quickly allowed researchers to zero in on the data drivers of the dependent variable under study.
The ability to integrate SAS fully into any analyses from JMP allows those analysts with access to both programs a new level of speed and flexibility while integrating the dynamic graphics of JMP with the breadth and depth of SAS’ analytic features. In addition, new scientific disciplines such as genomics, proteomics and chemometrics dictate the use of huge databases, which JMP 7 can handle. A 64-bit Windows or Linux system can now handle a nearly unlimited number of columns. The range of new features will be more fully addressed in my upcoming review. Now on to Mathematica….
The various blurbs at the Wolfram site strongly infer that this is far more than a new version. They claim that it is so radical an upgrade that an actual renaming of the package would be in order. Having just started using the new version, I couldn’t comment on the hyperbole, but the increase in speed alone is breathtaking. Even for a program that is noted for its ease in dealing with large numbers, the facility for complex calculations in ever-decreasing time periods is rather amazing (to a guy who learned most of his calculating in the pre-PC/hand calculator era, basically pen and paper).
The “full” review that appears in this issue does not begin to do it justice, and I strongly recommend viewing the Web site (www.wolfram.com/products/mathematica/newin6). Here the interested reader can see both a summary of new features and a detailed listing of new functions. During an introductory Webinar, it seems that, among all of the new features in Mathematica 6, the developers are most willing to highlight the adaptive graphics capabilities whereby an analysis may be quickly made into a dynamic graphic with suitable controls. As with JMP, the analyst can easily call up these visual manifestations of the equations, but now over the full range of Mathematica’s considerable range of functions.
As a long time laborer over many a mathematical proof and demonstration (at the sophomore level it would seem!) it was thrilling to see the FullSimplify command extended to allow automated proving with symbolic operators. As a statistician, the addition of many matrix and linear algebra functionalities was exciting. I have rarely used this software for routine statistics, but in the area of mathematical statistics, there is much to make life easier. Enough of the good, now on to the bad….
My minor rant this month concerns Microsoft’s new Vista operating system. There is actually much to appreciate in this new tool. However, once we can get by its paranoia with pseudo-security measures, analysts may have a major gripe in terms of running their standard software. A recent column in my local newspaper had the resident computer columnist loudly complaining that the new OS would not run older versions of his popular, non-technical software and that it seemed to be telling him, “Get an upgrade, cheapskate!” This philosophy is barely palatable for $50 to $100 programs, but for those in the $1,000 to $5,000, it is totally unreasonable.
Unfortunately that is what happened to this technical analyst. As I migrated to the new system, only half of the software would load and, of the half that wouldn’t, only two of the companies were working on compatibility patches for their older versions. I’m talking about either the newest versions that happened to come out before Vista or versions only one back from the latest and greatest. To me, this is almost as maddening as software that demands you register each installation! (Notice to software marketers: I may not be the only one objecting to this foul practice.)
John Wass is a statistician based in Chicago, IL. He may be reached at editor@ScientificComputing.com.