SAP Takes a Pass on SPSS

The business analytics boom has vendors searching for new ways to make themselves attractive to interested businesses. Recently SAP went so far as to reinvent the wheel when they ditched SPSS and created BusinessObjects Predictive Analysis, hoping to gain attention from the enterprise market. ;

Today Cindi Howson took an in-depth look into SAP’s BusinessObjects Predictive Analysis application. In soft launch style, the company has made the offering accessible to a number of its customers before general release.

This is a new direction for SAP as their current business model included reselling and embedding analytics software from SPSS, which was purchased in 2009 by BI competitor IBM. The move makes sense, as SAP would like to refrain from giving big blue a cut of their analytics earnings.

Business Objects Predictive Analysis utilizes the R programming language. R has been around since the 90’s, and recently gained traction in the business analytics market. Vendors including Information Builders, TIBCO and Oracle have all joined the statistical language’s bandwagon. Along with their implementation of R, SAP has included a GUI to complement the open source language.

The two main parts of the application focus on modeling and visualization. Howson notes the application has plenty of visualizations, however it was not developed to compete with the likes of Tableau or cannibalize SAP’s BusinessObjects Explorer. BusinessObjects Predictive Analysis is more of a product for statisticians as opposed to executives.  ;

Processing for the application takes advantage of SAP’s HANA in-memory database, which includes a statistical function library. In-database processing is a growing trend that can be tracked as far back as 2008 with SAS’s Teradata platform.

As data has grown, more vendors have turned to this model to reduce the time required for large data analysis. Along with SAS and SAP’s new offering, Netezza, EMC Greenplum, DB2, AsterData and Oracle all offer forms of in-database processing.

One of the speed bumps in advanced analytics is the process of creating models. The skillset required to develop these models is facing increasing demand, leaving data scientists in shorter supply. In an effort to ease the burden on their current and potential clients, vendors have begun shipping pre-fab models with their products. Howson mentions that SAS has the lead in this area, but IBM is gaining steam as well.

The new offering plays well with SAP’s BI platform, with the ability to read .UNV and .UNX model formats. Unfortunately it does not allow for simple exporting of results to dashboards or reports. While not specifically speaking to that issue, SAP has mentioned a focus to further integrate its analytics applications with BusinessObjects Predictive Analysis. One example given was the possibility of their Smart Meter application tuning its clustering and segmentation algorithms for energy consumption into their new offering.

It remains to be seen if this move will make SAP a stronger player in the market. However, moving away from SPSS along with this new application certainly provides new potential for their analytics platform.

Related Stories

SAP Dishes on Database Market Ambitions

New Path for SAP: In Memory Computing, Predictive Analysis Converge