If it’s true that the future will be won by the companies who leverage their information best, then SAP’s customers will be dancing in dollars.
Not only because of HANA's in-memory processing capabilities and the big data informed Customer Engagement Intelligence solutions SAP introduced last month, but also because the software giant now provides some of the world’s most powerful predictive analytics capabilities via the integration of its KXEN acquisition.
For those who aren’t familiar with KXEN, it’s a predictive analytics software company whose primary offering, KXEN InfiniteInsight, revolutionizes forecasting in two ways; first, it empowers business analysts with data scientist-like capabilities via automating much of their hard, messy and time-consuming work, and second, through the same automation, it can build more models with more variables (thousands versus hundreds) faster than large armies of data scientists.
Proof in Practice
While this certainly sounds good in theory, in practice it has yielded some pretty impressive results for the campaign to (re)elect U.S. President Obama and clients like AAA, Sears, Lowes, Shutterfly, Rhapsody (predicting what song you’d like to hear next) and Cox Communications among many, many others.
It’s worth noting that these results weren't just hearsay to SAP when it was looking to add analytics muscle to its portfolio; the companies shared a number of clients who used their products side by side.
Goodbye KXEN, Hello SAP InfiniteInsight
Beginning today the KXEN name is going away and the predictive analytics product will now be sold by SAP as SAP InfiniteInsight. It will also be available bundled with SAP Lumira for visualization, and SAP Predictive Analysis.
Faster, Simpler, More Powerful, Less Error
SAP InfiniteInsight offers a series of components. Some of them include:
- The Explorer -- It allows power users to define a broad set of reusable business components, called analytical records, which can be applied over and over again to automatically create the analytical data sets used for modeling. SAP says that this approach is orders of magnitude faster and results in far less human error than traditional handcrafted techniques. It also opens the door for business users to self-service their predictive modeling needs, while guaranteeing accurate results.
- The Modeler – Business analysts can stop fretting over complicated math or deciding which algorithm to use. InfiniteInsight automates the building of sophisticated predictive models for every data mining function and helps users quickly get to the right algorithm for their business problem with a model built for accuracy and optimal results.
- The Scorer -- Once a model has been built, it needs to be deployed. The Scorer deploys optimized scoring equations directly in database to SAP HANA® and other major database platforms.
- The Factory -- One “what if” question is never enough, but asking too many of them can be problematic. With traditional predictive methods, it’s impossible to optimize great numbers of campaigns, to become overly granular in targeting, or to retrain models on a regular basis. It takes too long and competes with other business priorities. InfiniteInsight Factory eliminates these limits. It is industrialized modeling with a factory approach so that you can build, deploy and improve any number of predictive models.
Big Data: Many Questions, Answers Needed Now
Before InfiniteInsight belonged to SAP, SAP’s 360 Customer solution and KXEN InfiniteInsight worked side by side at Eldorado, which some call the Best Buy of Russia. The consumer electronics and households appliance retailer runs more than 1.5 million point-of-sale transactions spanning 420 product groups and 8,000 products each month.
Data generated from those activities is used to create forecasts for pricing and promotion analysis, store clustering, store location selection and sales and purchasing planning.
From an analytics perspective, the company runs more than 500 models per month via InfiniteInsight. If these models had to be constructed manually they would take weeks or months to build and require a huge (50 plus) army of data scientists working around the clock 24/7 and that still wouldn’t be sufficient.
The combination of SAP and InfiniteInsight is. It has allowed them to build and train enough models using enough variables to generate a 10 percent improvement in forecast accuracy.
We’ve Only Just Begun
SAP has big plans for InfiniteInsight which will eventually include integration with HANA. It will be interesting to see if the marriage of these technologies will produce an unstoppable “super power.”