Welcome to the new SAP Predictive Analytics 2.0! SAP quietly released the new generation of their user-facing predictive analytics offering in February 2015. This represents the first time their 2 separate tools, Infinite Insight (formerly KXEN) and Predictive Analysis, have been shipped together. This release also introduces new names for both products, as well as some of the sub components that were part of the former Infinite Insight tool. The table below shows a translation of the new names and the equivalent previous names. Despite the name changes, there is relatively little new functionality in either of the tools with the 2.0 release. Unified Installer and Launch Screen Probably the most notable change for the 2.0 product is that there is now a unified installer, which installs both the former Infinite Insight (KXEN) tool as well as the SAPPredictive Analysis tool. Due to the complexity of the Infinite Insight architecture, with 2 versions (server and desktop) and 2 components for installation in the server version, this causes additional complexity when selecting installation packages. All of the installation packages include the same copy of the Expert Analytics tools (previously SAP Predictive Analysis). The screenshot below shows the different installation package options. The most common installation package is the desktop package, which includes the full version of Expert Analytics and the local version of Automated Analytics.
In order to install any of the above packages, any existing copies of Predictive Analysis and Lumira will have to be uninstalled before any of the 2.0 installation packages will run. Luckily any data saved in the Lumira and Predictive Analysis content folders is preserved and is accessible through the new installation. After installing Predictive Analytics 2.0 and starting up the program, the screen below appears, which closely resembles the previous Infinite Insight launch screen, though it has been re-branded to reflect the new Predictive Analytics banner. Clicking on the Expert Analytics section reveals the screen below, with a link to launch the Expert Analytics application.
While it is somewhat convenient and time-saving to download one installation package to install both programs, that time savings is quickly erased by the increased time to open the application. Specifically, the unified launch screen requires first starting up the Automated Analytics program (this takes 30-90 seconds on my machine), and then to launch the Expert Analytics program, which takes at least an additional 1-2 minutes on my system. This seems excessively confusing for a user that wants to jump directly into the Expert Analytics tool and does not want to use the Automated features. Additionally, while SAP has unified the installation process and the license for each tool includes the other automatically, the licensing for both products is not unified. The Automated Analytics tool licensing still requires a license config file that typically ships with the installation while the Expert Analytics tool requires a license key to be entered at installation like other SAP tools. Despite the Predictive Analytics 2.0 branding, Expert Analytics and Automated Analytics are clearly still two entirely separate tools and offer completely different user experiences. Enhancements to Expert Analytics Most of the functional enhancements that were released with the initial 2.0 release are on the Expert Analytics side, which is interesting, as it may suggest that SAP has decided on folding the Infinite Insight toolset into the Lumira codeline rather than continuing the Infinite Insight tool. Access data from SAP BW With this release, Expert Analytics can now access data from SAP Business Warehouse BEx Queries and InfoProviders (with or without variables) directly and use this data in Offline mode. While the Business Objects Universe integration options have been available since the earliest releases of Predictive
Analysis, many BW users have been left feeling underserved, with the only option to extract datasets to a file or build a universe on top of the DSO to access BW data. This enhancement furthers SAP s integration advantage over all other predictive tools and further enhances the value of the tool to organizations with a mature SAP business intelligence environment. Share Lumira Content via Lumira Edge Server and SAP BI integration Predictive Analysis has also been enhanced to facilitate opening content located on the new Lumira Edge Server and Business Objects environments. Content from either of those locations is available in the menu on the main screen of Expert Analytics (shown below) and the Share tab now has the option to upload Lumira content to these locations as well. This is for Lumira content only, and visualizations on predictive datasets cannot be shared.
In order to save information to Lumira Edge, you must File Save As and select the Lumira Server location. Lumira Server, Lumira Edge, and SAP BI connections must be configured prior to use under File Preferences Network within Expert Analytics. Introducing the Automated Prediction Library (APL) to SAP HANA Along with the launch of the first unified user-facing tool, SAP has released the Automated Predictive Library (APL), an addition to the existing Application Function Libraries component of HANA (which also contains the Predictive Analysis Library (PAL) and the Business Function Library (BFL)). The APL contains 3 of the Infinite Insight algorithms optimized to run entirely on HANA. Currently the APL supports Classification/Regression, Segmentation, and Time Series predictive models, as well as cutting (sampling) strategies required for the prior 3 models. The APL must be installed manually on the HANA instance and is only available for HANA SPS09 and later. For full details on installing, granting security access for, and using the APL, refer to the Automated Predictive Library reference guide. APL algorithms supported by Expert Analytics are listed below:
The screen below is the configuration window for the HANA Auto-Classification algorithm, which allows the user to select inputs and a predicted (target) column, must like the other classification algorithm configuration screens. See future blog posts for additional information on the APL. Hillary Bliss, Business Intelligence Consultant Decision First Technologies Hillary.bliss@decisionfirst.com twitter @HillaryBlissDFT Hillary Bliss is a Business Intelligence Consultant specializing in data warehouse design, ETL development, statistical analysis, and predictive modeling. Hillary works with clients and vendors to integrate business analysis and predictive modeling solutions into data warehouses based on their data and business needs. With Decision First Technologies, Hillary uses Data Services, Web Intelligence, Predictive Analysis, and HANA. Hillary has Masters in Statistics and an MBA from Georgia Tech.