Jason Verlen and Bernard Spang took one of the breakout sessions on analytic innovations. IBM, as noted yesterday, sees a progression of analytics from descriptive to predictive, prescriptive and ultimately cognitive. The purpose, of course, of big data and analytics is to drive better decisions. Managing big data is a necessary step but acts only as a foundation for deriving smarter analytics and these, in turn, are only useful if they can be used to improve decision-making. As Jason says these must be “pushed to the glass” either to influence a decision-making in a decision support system say or to drive an operational system to respond in a decision management system.
He level-set with a number of widely held beliefs around big data that are basically totally wrong – that Big data means Hadoop, that all older systems will have to be replaced in an era of Big Data, that only trendy companies need Big Data or that transaction data doesn’t matter any more. What Big Data does do, however, is give organizations new opportunities for analytic insight by bringing new data types together with existing, structured data.
To help organizations with Big Data IBM has a set of solutions, centered around the PureData and InfoSphere products. Yesterday we covered some of the new products (DB2 BLU Acceleration, Big Data Platform and PureData System for Hadoop). Plus there is the SPSS and Cognos stacks to develop analytics and deploy those analytics using Decision Management and signature solutions. These can be thought of as supporting a set of use cases:
- Business Intelligence
Cognos BI dynamic cubing and DB2 BLU offer big performance improvements for BI professionals while adding flexibility by avoiding the need for pre-aggregating. - Social Media
The new social media platform (also discussed yesterday) is built on InfoSphere BigInsights. This is now available on-premise and i the IBM SmartCloud. - Decision Management
IBM focuses on Decision Management because everyone wants to apply predictive analytics but skills are hard to come by. By making it easier to build, deploy and use predictive analytics in operations IBM helps address this. As regular readers know Decision Management combines business rules, predictive analytics and optimization to improve high volume, operational, real-time decisions by giving front-line staff recommendations or actions. SPSS Analytical Decision Management is often used for non-knowledge makers, delivering actions or recommendations to front-line workers. A wide range of preconfigured solutons exist from IBM from Next Best Action to Predictive Debt Collection. - Predictive Analytics
Building predictive analytics (e.g. for deployment in Decision Management Systems) is supported by SPSS Modeler for expert users while intermediate users can take advantage of automation of many of the detailed steps that an expert might do by hand. SPSS Modeler Advantage delivers a point-and-click interface to build models. Under the covers this shares the same environment so experts and beginners can share the model building logic. Modeler takes advantage of the new BLU acceleration as part of its in-database capabilities to build models faster. It’s integrated with InfoSphere Streams so that scores and decisions can be made against data in motion. Finally there is new entity analytics (blogged about here in my recent update on SPSS Modeler). - Predictive Discovery
This is where the new product, SPSS Analytic Catalyst comes in. Big Data is complex and people are trying to apply more statistical approaches to their data discovery. Catalyst is designed to apply core statistical approaches to a dataset automatically and then brings back a set of top insights along with explanations of them. It provides top drivers as well as top associations between fields in your dataset. At any point the user can drill into the results presented, into the models under the covers, see what’s significant, favorite an analysis etc. Catalyst is built on Analytic Server and runs on desktops or tablets. Analytic Server can support SPSS Modeler also and provides support for R.
Cognos dashboards can bring all this together, combining the social media analysis, traditional BI and reporting elements, decision management outputs etc. All designed to bring Big Data Analytics to the point of impact.
Comments on this entry are closed.