We got a quick overview of SPSS Decision Management (see my post on SPSS DM 6 or indeed the whole blog for my view of this). SPSS Decision Management combines predictive analytics, business rules and optimization/simulation to optimize high volume decisions like “should I search this car at the border” or “should I investigate this claim” or “is it worth trying to save this customer”. SPSS Decision Management restricts freedom of action with respect to building analytics so that more people can more effectively deploy analytics. 7 steps are the basis for the product: Select data; Selections/Exclusions; Define allowed outcomes; Define rules and select existing models or create new models; Combine rules and analytics and assess impact to optimize outcomes; Deploy; Report. These steps are presented through pre-configured user interfaces focused on a particular solution area.
IBM says that SPSS Decision Management is showing great momentum, in particular it is being used very successfully by IBM to position predictive analytics overall. The solution focus and deployment management of SPSS Decision Management helps focus conversations about predictive analytics and SPSS DM 6 has been sold successfully in industries as diverse as insurance, retail, telco, government, automatic, banking and pharmaceuticals. Selling SPSS DM 6 also pulls SPSS Modeler and modeling consulting as well as positioning companies for the kind of successful project that leads to follow on projects.
From a roadmap perspective SPSS has been delivering on various items – 3 from last year’s plan, 1 new one:
- More blueprints for new solution areas
Began with customer treatment and claims fraud. Now adding retail promotions and transactional fraud detection (for banking)
- Integration with the ILOG business rules management system
Later this year SPSS DM 6.2 will allow you to use ILOG rulesets instead of authoring the rules directly in SPSS Decision Management. This allows existing ILOG customers to use their rulesets in the product and to allow customers who outgrow the rule management capabilities of SPSS Decision Management to move to ILOG for long term/enterprise rule management.
- Integration with Cognos reporting
Cognos BI 10.1 can be used both as a data source (allowing you to build models against a data source defined in Cognos) and as a target for scoring (so predictive analytics can be pushed back into Cognos reports)
- Integration with social media
In addition there is integration work with the new Customer Insight product. Clearly the growth in social media has both made the purchase behavior of many consumers more complex while at the same time providing new information about consumers and their thinking. If this social data can be analyzed, this insight can be validated with surveys and combined with internal data to build more accurate predictive models and drive better decisions using SPSS Decision Management. You can also use social media data to define segments of anonymous users and then look for similar segments driven by internal data to make even anonymous social media data actionable.
- Integration with Netezza
Currently have SQL pushback (deploying models as SQL) but moving rapidly to true in-database analytics that takes advantage of the Netezza Twinfin data mining algorithms to build models in-database.
Futures probably include more and better integration with business rules, use of optimization to improve decisions, more pre-packaged solutions. Decision Management is clearly an emerging and broad market for IBM so expect more from IBM in this area.