Predictive Analytics: A Shopper's Checklist
A new report offers buyers a baseline to use as they shop for general predictive analytic technologies.
Analytics are all the rage. So-called "advanced analytics" are even hotter. Although the rise of "advanced analytics" as marketing spin invites skepticism -- it's not always clear what qualities, categories, or metrics qualify as "advanced" -- most BI professionals agree that predictive analytic technologies comprise a kind of advanced analytics in their own right.
Of course, distinguishing "advanced" (or mature) predictive analytic offerings from the rest can likewise be difficult. Established players such as SAS Institute Inc. and IBM Corp. subsidiary SPSS Inc. can lay claim to predictive analytic bragging rights, while relative newcomers such as SAP BusinessObjects, Information Builders Inc. (IBI), and Oracle Corp. (among others) are looking to distinguish their products. It can make for an exercise in RFP frustration.
A new report from market watcher Forrester Research gives buyers a baseline to use as they shop for general predictive analytic technologies.
For starters, writes Forrester analyst (and industry veteran) James Kobielus, a viable predictive analytic offering needs to satisfy a number of non-technological criteria. Kobielus cites feature/functionality, market presence, and application architecture as three key differentiators in kind.
Mature or viable predictive analytic entrants share several core features, according to Kobielus:
- A built-in ability to discover, extract, profile, transform, cleanse, or perform other (data integration-like) operations on source data
- An ability to "develop and validate multivariate predictive models" using a range of different methods, including classification, regression, and scoring
- A robust front-end interface that facilitates interactive engagement with and exploration of predictive results. (This front-end should permit analysts to "visualize, explore, and manipulate" predictive models, as well as to perform common kinds of analysis, including forecasting, simulation, and what-if types.)
Kobielus cites an additional functionality sine qua non: an ability to both store and manage analytical data sets in a centralized repository of some kind, such as an enterprise data warehouse (EDW) or a data mart.
Architecturally, Forrester limited its assessment to predictive analytic technologies that are cross-domain as opposed to domain-specific. Forrester also surveyed only vendors that had at least 50 customers and that generated at least $5 million in sales of predictive analytic software licenses, maintenance contracts, or subscriptions.
Not surprisingly, traditional powers such as SAS and SPSS still sit top the predictive analytic market, Forrester says. "The [predictive analytic and data mining] leaders have supported large enterprise customers' advanced analytics needs for many years. All offer mature, high-performance, scalable, flexible, and robust [predictive analytic and data mining] solutions that combine a wide range of statistical algorithms with integrated support for in-database analytics and a broad range of information types," Kobielus writes.
What's surprising, however, is the showing of some of the predictive-analytic-come-lately vendors. Although players such as SAS, SPSS, KXEN Inc., Oracle Corp., and Portrait Software field the most functional or respected offerings, Forrester singled out late-comers such as Fair Isaac Corp. (FICO), Angoss Software Corp., and TIBCO Software Inc. as creditable competitors.
"[These players] are aggressively challenging the [market] Leaders with mature, fast-evolving solution portfolios that have gained significant adoption among large enterprises in many verticals," Kobielus writes. "Just as important, these Strong Performers have established themselves as innovators with functionality in such key [predictive analytic and data mining] areas as wizard-driven development automation, multi-business scenario modeling, interactive visualization, content analytics, sentiment analysis, social network
analysis, in-stream analytics, and open-source modeling languages."
In short, he argues, it's an awfully big predictive analytic tent.
SAS, SPSS, KXEN, and other vendors enjoy notoriously loyal customer bases, but predictive analytic latecomers can likewise point to "substantial" and "loyal" customer references. To Kobielus, this suggests that there's "plenty of opportunity for well-differentiated [predictive analytic and data mining] solutions" in both the mid-market and large enterprise.