Yesterday SAP’s Charles Gadalla conducted an SCN Analytics webcast on Predictive Analytics.
I think the best part of this webcast was the question & answer.
He said the usual disclaimer applies that things in the future are subject to change. My disclaimer – these are my notes, based on what I heard.
Charles said businesses are paying attention to advanced analytics and want things done in “real-time”
Figure 1: Source: SAP
Charles said the product went GA in December 2012
He said Forrester moved the product to the leader quadrant in Q1
Businesses are paying attention to advanced analytics and want things done in real-time
Figure 2: Source: SAP
Figure 2 shows Mitsui, a Japanese company, using HANA and R to speed up analysis and investigate tumors
Charles said HANA several times faster than traditional based systems
Figure 3: Source: SAP
Figure 3 review R, which is free, and used in areas such as banking, high tech
Figure 4: Source: SAP
Top left of Figure 4 is for business users; Predictive Analysis is for the business user.
PAL uses a C++ library
Figure 5: Source: SAP
Coming in the middle are planned innovations, including PAL support, PAL library, algorithm SDK – actual SDKs – bring in your own or use yours
He said that different personas being targeted
He said KXEN is like a blackbox, identify what looking for; the acquisition strategy has been around business user/persona
Question and Answer
Q: How is different from Lumira?
A: Lumira is a subset of Predictive
Missing Predictive tab in Lumira
Gain predictive tab with Predictive Analysis
Q: Where is Predictive going?
A: Business user is paramount, advancement is key
Q: How is this different from Data Mining
A: Data mining is looking at past data, predictive – turn equation around, interesting relationships – regression to look at the future
Q: Where does APO fit?
A: APO relies on where to fit demand
Looking at SAP fraud management, demand signal management, PLM SRM coming soon
See real value to have the person using APO to drive the forecasting and predict without leaving APO
Q: Is SAP building industry specific models?
A: Yes, this is coming through templates, best practices
Q: Can you only build PAL models or our own?
A: Going forward there will be a SDK to convert to C++
Q: Who builds math model – business or IT?
A: Start somewhere – he said “get ducks in water so start swim” – I don’t recall the answer
Q: Does PA support Text Analysis?
A: Supported today in HANA
Q: Can you give an example of how to do PA against loose data relationship? Category management?
A: Tide versus grapefruit scented Tide – not a linear relationship – classification, neural nets
Holistic framework, flat file – offer with Lumira with a flat file
Q: How ensure support if deploy a solution based on R?
A: R is open
Q: Is BW supported?
A: Inside of BW on HANA, look at analytic view, and compare on a flattened view
Customers want to connect to BW – this will happen
Q: Data prep for PA?
A: Turning to KXEN – auto binning and variable reduction – provides the data prepping
Data services will help
Q: Can you transfer from the cloud, KPI?
A: Yes, not limit to data warehouse, use PA/Lumira Cloud integration, KXEN also provides cloud integration
Upcoming SCN Analytics webinars, open to everyone:
SAP BusinessObjects BI 4.1 Upgrade
October 8, 2013
Empower Your Business By Creating, Visualizing, and Sharing Your Data
October 15, 2013