SAP Steps Up Commitment To Predictive Analytics

SAP BusinessObjects Predictive Analysis will integrate with Hana, take on SAS and IBM SPSS modeling and analysis tools.

Cindi Howson, Founder, BI Scorecard

March 15, 2012

4 Min Read

SAP last month quietly "ramped up" SAP BusinessObjects Predictive Analysis. Ramp up is the vendor’s approach to releasing production software to a limited number of customers. The product is expected to be generally available later this year.

Predictive analytics has become an increasingly important part of the larger business intelligence (BI) market and an area in which SAP has lagged behind chief competitors IBM and SAS. SAP currently resells and embeds analytics software sourced from SPSS, but with IBM’s acquisition of SPSS in 2009, it seemed only a matter of time before SAP would develop an alternative.

SAP BusinessObjects Predictive Analysis, the company's new product, has been shaped by competitive and market forces, including the momentum of open source R, in-memory and in-database processing, and the convergence of analytics with BI.

Open source R: This statistical language emerged from a project initiated by academics in New Zealand in the mid 1990s. A number of vendors, including SAS, have added support for R as it has gained popularity. Information Builders WebFocus Rstat and TIBCO Spotfire S+, for example, are both based on R. SAP BusinessObjects Predictive Analysis provides a graphical user interface to R.

[ Want more on BI? Read 6 Predictions For Business Intelligence In 2012. ]

SAP BusinessObjects Predictive Analysis has two components: model development and execution, and data visualization. I was impressed by the number of visualizations Predictive Analysis automatically generated to identify patterns in data. To be clear, this is not a visual discovery tool for business users that competes with the likes of SAP BusinessObjects Explorer or Tableau Software. It's a tool for statisticians.

In-database processing: Statisticians have typically done their analyses and model building on data that resides in flat files or in proprietary data-set formats. As data volumes have grown, database and analytic vendors have been pushing more of the processing into the database, where analyses of large data sets can be completed much more quickly.

SAS was one of the first to support in-database processing, working with Teradata in 2008, and it now supports the approach in Netezza, EMC Greenplum, DB2, and AsterData databases as well. Oracle last month released Oracle Advanced Analytics, which bundles Oracle R Enterprise software and supports in-database processing of R models inside the Oracle database. Predictive Analysis pushes processing into Hana, SAP’s in-memory database, which has its own statistical function library.

Mainstream analytics: There has never been a cry for mainstream analytics in the way there have been calls for democratizing business intelligence. Developing analytic models remains a task for skilled statisticians, but there have been calls for improved integration and easy consumption of model-derived data. Some BI vendors have made it easy to call on models and transparently embed the results within a dashboard or report. An example is a report that displays sales by customer, with a model-derived predictive flag to indicate the likelihood of churning.

MicroStrategy was one of the first vendors to provide such seamless integration, but adoption has been slow. There has been greater demand for pre-built analytic applications, a market in which SAS is the leader and IBM is growing.

SAP BusinessObjects Predictive Analysis is well integrated with SAP's BI platform in that the tool can access a Universe (data model) in either the version 3 (.UNV) format or the newer version 4 (.UNX) format. The level of integration is similar to what SAP provides for SPSS software. However, there's currently no easy way to embed model results in a dashboard or report. SAP says its various analytic applications will take advantage of Predictive Analysis modeling capabilities over time. For example, the SAP Smart Meter Analytics application uses clustering and segmentation algorithms on energy consumption that could be developed and tuned in Predictive Analysis.

SAP BusinessObjects Predictive Analysis is not the vendor’s first entry into advanced analytics, but it represents a bigger commitment than earlier efforts. It’s too early to say whether it will have any impact on SAS or IBM SPSS market share. However, it certainly will improve the capabilities of SAP analytic applications and it will put SAP on the short list of vendors to investigate for any customer that is new to predictive analysis.

Cindi Howson is the founder of BI Scorecard , an independent analyst firm that advises companies on BI tool strategies and offers in-depth business intelligence product reviews.

About the Author(s)

Cindi Howson

Founder, BI Scorecard

Cindi Howson is the founder of BI Scorecard, a resource for in-depth BI product reviews based on exclusive hands-on testing. She has been advising clients on BI tool strategies and selections for more than 20 years. She is the author of Successful Business Intelligence: Unlock the Value of BI and Big Data and SAP Business Objects BI 4.0: The Complete Reference. She is a faculty member of The Data Warehousing Institute (TDWI) and a contributing expert to InformationWeek. Before founding BI Scorecard, she was a manager at Deloitte & Touche and a BI standards leader for a Fortune 500 company. She has been quoted in The Wall Street Journal, the Irish Times, Forbes, and Business Week. She has an MBA from Rice University.

Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.

You May Also Like

More Insights