Convergence : A Catalyst for Fuelling Democratization of Predictive Modelling
This year is going to witness two significant innovations coming together to help fuel the democratization of predictive modelling. SAP’s acquisition of KXEN in 2013 brought about a new dimension in the way SAP planned to position predictive capabilities with our customers. SAP InfiniteInsight was the reincarnation of KXEN in the SAP world. At the time of the acquisition, SAP had a home-grown predictive analysis product. SAP Predictive Analysis was a product focused on the data scientist persona (aka. expert users) who has an in-depth knowledge of the algorithms and the predictive modelling process.
These two products are now coming together into the next generation product – SAP Predictive Analytics
The first deliverable, as part of the converged product, is the single installer for both SAP Predictive Analysis and SAP InfiniteInsight, which released to customers today. Expert analytics and automated analytics are the terms used for SAP Predictive Analysis and SAP InfiniteInsight respectively in the converged product. This enables the user to install a single product and have access to both modelling environments.
Going forward, SAP Predictive Analytics will bring together a seamless modelling experience (for both business and expert users) and the benefits of operationalizing the predictive models.
- For business applications that are capable of using and consuming SAP HANA services, be it on premise or in the cloud, SAP InfiniteInsight’s technology (algorithms) is being made available inside SAP HANA as an automated predictive library (APL).
- To enable integration scenarios into either SAP or third-party cloud applications, the predictive algorithms are also being delivered as a set of machine learning services in the cloud. Please note this is currently available for internal usage only.
- The machine-learning cloud services will be made available on SAP HANA Cloud Platform, and is planned to be available in the next release of SAP Predictive Analytics.
- Several algorithms have already been made available as APL and machine learning services, and are being embedded into SAP business applications like:
- SAP Cloud for Customers for prospect scoring
- SAP Simple Finance (specifically for cash management)
- SAP Customer Engagement Initiative for campaign targeting and product recommendation
- SAP Smart Business cockpit for displaying key performance indicators driven by the solution
- SAP Predictive Maintenance for predicting machine failures to facilitate targeted maintenance
- SAP Predictive Analytics also offers APL nodes in expert analytics and Big Data support via Hadoop and Hive
SAP Predictive Analytics came about because there were several benefits that our end-users enjoyed in both the predictive products from SAP and KXEN.KXEN’s solution offered capabilities for easy consumption of predictive modelling to the business users who aren’t expected to be experts in predictive modelling. It also offered several ways to operationalize predictive models. The eventual SAP InfiniteInsight offered many features including:
- An easy automated modelling capability
- A rich interface to deploy models into the production and real time monitoring
- The ability to create analytical data sets using Explorer functionality, SAP InfiniteInsight Recommendation and others
SAP Predictive Analysis focused on the data scientist (expert user) who has an in-depth knowledge of the algorithms and the predictive modelling process. Features included:
- A modelling environment that enables an easy way to build complex models in both offline and SAP HANA online modes
- In case of offline mode, the user can build complex models using several “R” algorithms packaged along with the product, whereas in case of the SAP HANA online mode the user has access to optimized predictive algorithms as part of the predictive analytics library (PAL) which are packaged along with the product.
- A simplified interface for the user to get any “R” algorithms or PAL algorithms as reusable components that aren’t currently packaged as part of the product and use them in conjunction with the existing algorithms to build complex models and publish the models as stored procedures / views on SAP HANA for consumption.
Now that the two products have come together into the next generation product – SAP Predictive Analytics – the customer will find the best of both the products with features that will enable them to embed predictive in a pervasive manner and to a much wider spectrum of applications. In other words, the converged product will act as a catalyst for fuelling the democratization of predictive modelling.
For more information visit www.sap.com/predictive.