TDWI | Training & Research | Business Intelligence, Analytics, Big Data, Data Warehousing

SPSS 15: Has Bells and Whistles, Will Go Mainstream?

Officials say the new suite will help SPSS break through to mainstream success.

• By Stephen Swoyer
• September 20, 2006

Billion dollar power-player SAS Institute Inc. might get most of the ink, but it’s not the only number-cruncher with business intelligence (BI) aspirations. Consider statistical analysis and predictive analytics stalwart SPSS Inc., which—with a pedigree that’s nearly as venerable as that of its Cary, N.C.-based competitor—is also enjoying fresh success in the BI market space.

Buoyed by new market research from International Data Corp. (IDC), whose 2006 BI market forecast has it as an up-and-comer (with strong year-over-year growth) in the fiercely competitive enterprise BI segment, SPSS this week unveiled a new version of its core BI and data management suite, SPSS 15. Boasting improved usability features, enhanced report formatting capabilities, strengthened support for scripting and script-driven transformations (as well as support for Python and other scripting engines), SPSS officials say SPSS 15 will help take their company mainstream.

“This is where you’ll see [SPSS] going from more of a desktop tool really to more of an enterprise-level application,” says Kyle Weeks, senior product marketing manager with SPSS. “For example, we align SPSS 15 with some of our other enterprise products, such as Clementine and Predictive Enterprise Services, [and] as they become more and more a central part of the way people do biz, these really become more and more corporate assets then. Like any kind of corporate asset, they have a certain set of needs then, they need to be locked down, standardized, controlled, and versioned.”

And that, Weeks says, is just what SPSS has attempted to do, especially over the course of its last three platform releases. The good news, market watchers say, is that the company’s investments appear to be paying off.

“SPSS is the second-largest advanced analytics vendor. Its focus on predictive analytics paid off in 2004 and 2005 after several years of lower-than-market growth rates. In many cases, SPSS has also been able to cross-sell its query, reporting, and analysis tools into its base of advanced analytics customers,” wrote analyst Dan Vesset in IDC’s recent 2005 BI market share report.

Enter SPSS 15, which debuted this Monday. Key to the new release are enhanced data management capabilities, plus improved reporting functionality, including extended support for SPSS’ Chart Builder drag-and-drop chart authoring interface; new chart types (such as dual-Y axis and overlay charts); and a PDF export option. SPSS 15 ships with a new Export to Database Wizard that lets users easily write back to databases from within SPSS. Elsewhere on the data management front, customers can tap “Custom Attributes” to use their own dictionary information for variables; work more easily with very wide data files by customizing variable sets; use algorithms designed for nominal attributes—such as Naïve Bayes and logit models—with Optimal Binning in the SPSS Data Preparation add-on module.

SPSS 15 also boasts improvements to its bread-and-butter number crunching capabilities, with support for generalized linear models (GZLMs) and generalized estimating equations (GEEs), via its SPSS Advanced Models add-on module. Similarly, SPSS 15 enables customers to predict ordinal outcomes from a complex sample design via its SPSS Complex Samples add-on module.

One of the biggest enhancements in SPSS 15 builds on technology—i.e., the SPSS Programmability Extension—which SPSS introduced in version 14 of its software. This lets users tap industry- or open-standard programming or scripting tools (such as Visual Basic or Python) to build automated jobs that can be chained together, and (for users of Clementine or other SPSS tools) are able to hand off output (and script any necessary transformations) from one tool to another. While the SPSS Programmability Extension facility debuted last year in SPSS 14, Weeks explains, the new improvements let users create procedures in external languages and build them into SPSS (complete with user interface support and an interactive pivot table, via SPSS’ Output Viewer). Programmers can now read and write case data and create new variables directly, too.

“We started that with SPSS 14 and we take that to the next level in SPSS 15,” Weeks maintains. “Really, our plan is to make it possible for [customers] to embed the entire analytic guts of SPSS in their applications, so whether it’s Python or a .NET application, they can use that [the SPSS Programmability Extension] to actually drive SPSS from within their applications.”

Nor is this an SPSS-only play, Weeks insists. “SPSS has something called the Output Management System, where the output doesn’t have to go to our proprietary Output Viewer, it can spit out XML, for example,” he comments.

IDC’s Vesset isn’t the only one who gives SPSS high marks for executing on its BI go-to-market strategy. King Douglas, a senior analyst with American Airlines Inc. and a long-time SPSS user, says he likes the direction in which SPSS is going—especially in its last three releases. “From versions 13, 14, and now 15, they’ve really gone back and made a star of SPSS, before that for a while it was stuck in a byproduct of its own success,” Douglas comments. “I think [version] 15 is the best release yet. One of the things I really like is the [support for the] exportation of SPSS output into Web pages, or [support for] export out into PDF documents, but they also include hyperlinks.”

Douglas also touts the Python integration, which he says eliminates the need for proprietary SPSS scripts. In the old model, he says, “you could manipulate output after it was done—there were macros and various types of bells and whistles where you could manipulate data and certain objects, but there was a main flaw, which was that if you tried to launch a script to manipulate data, it ran independently from SPSS, so if I was running a large series of tables and analyses, it may run independently, but there’s no sequence to my programming. So you would have to tell yourself to stop here, run the script, when the script was done, start again,” he explains. Not so with the new Python scripting support, Douglas avers: “The older system was just much, much harder to use and didn’t run synchronously with SPSS.”