SAP presentation at the Chief Analytics Officer, Fall 2016

1. Data Science Automation: The Next Frontier Doug Freud Associate Vice President, Innovation, Data Science, & Data Strategy
2. If you do not wake up everyday and think about how your company is going to disrupt your industry you can bet that one of your competitors is already doing it.
3. Amazon’s Digital Disruption of Retail World
4. In the Beginning……
5. 15 years ago A/B tests ruled! Amazon.com’s high upsell cart Add an item to your cart and get a whole page of product recommendations SVP at the time said, “No way. Do not test it. Anything that distracts shoppers from checkout will hurt our conversion rate.” Greg Linden said, “Okay I’ll test it.” Generated $7m attributable sales a day in 2004. First Generation Recommendation Engine
6. Today the Recommendation Engine Accounts for 35% of Sales Revenue
7. ML is the New BI BI: Powered by humans for human consumption. Understanding what has happened ExploratoryAnalysis: Powered by advanced techniques (Statistical/Machine Learning/Data Mining) for human consumption. Ad- Hoc / R&D approach. Why did it happen and how to improve and optimize ML: Machines/Systems are making decisions or recommendations that are powered by ML algorithms. The probabilities or rules are consumed by software or processes, and automatically continue to learn and improve.
8. “Interconnected” Data is Growing Exponentially *** Gartner says the “The Internet of Things” install base will grow to 26 billion units by 2020 ** How Much Data Is Generated Every Minute On Social Media? 26B Connected things by 2020*** 3.5 BGoogle searchper day* “Data is the new oil, but in order to monetize it requires predictive analytics” IOT • expected to be a $1.9 billion market by 2020, according to Gartner. • 90 percent of all data generated by IOT is never analyzed or utilized in business decision processes. • Sensors & Computer logs Consumer Internet • Internet Surfing, email, Social media • 1.4 Billion active Facebook users, Twitter 21 million tweets per hour, 300 Million active Instagram users, & 290,728 Tinder matches per minute • Text & Web Logs
9. Challenges for the Chief Analytics Officer Shortage of Analytical Talent Complicated Data Management Shorter Decision Cycles
10. Keeping Pace Requires Automation: A Factory Approach Models Per Month Source: Adopted from Factory Analysis vs. Craftsman Analysis, Gartner, 2010 2005 2010 2015 2020 Factory Automation: • Creating Analytical Data Sets • Automation around Model Building • Deployment • Model Management
11. works with Lucy to make more accurate expert models works with Frederick to gather the data he needs. works with Susan on data mining tasks. ETL Champion Predictive Model Manager Platform Architect Data Scientist Sets business requirements and sponsors project Data Science is a Team Sport
12. Dynamically Creating Analytical Data Sets
13. Automating Modeling Process
14. Automating Model Management
15. 17© 2016 SAP AG or an SAP affiliate company. All rights reserved. Fleet Level Information Machine Level Information PdMS Application shows insights about health score of machines and business information ERP + MRS Maintenance Schedules & Financial Information Action ~10% Reduction in Maintenance Costs
16. Machine Learning for SAP’s digital core, cloud and networks 18 • CV matching • Freelancer matching • EE scoring with NLP • Employee Lifetime Value • Career path recommender • Employee Churn Prediction • Service Finder • Decision Support • Predictive Maintenance • Smart Forecasting Model • Failure Detection and Prediction • Maintenance Planning Optimization • Invoice matching • Cash flow forecasting • Travel virtual agent • Vendor matching • Pay Me Now solution • Procurement recommenders • Semantic product understanding • Procurement risk diversification • Product recommender • Customer Lifetime Value • Social sentiment analysis • Automated sales forecasting • Cart abandonment prediction • Marketing action recommender 27 INDUSTRIES • SAP Big Data Margin Assurance • XM Exchange Media Video Advertising recommender SAP HANA CLOUD PLATFORM • Clustering • Regression • Classification • Anomaly detection • Reinforcement learning
17. © 2016 SAP SE or an SAP affiliate company. All rights reserved. 19Internal A B C The Predictive Enterprise The Predictive Factory Improve bottom line through managing 000’s of data sets and models Predictive IP Generic and specialized algorithms Big Data Make economical sense of more data collected Embedded in Processes and Apps Cloud Easy to consume apps and micro services On Premise In tandem with SAP HANA and SAP BI P </ > 10101010 00100110 0111 High and Low Touch UX’s For all data scientists: citizens and experts Functional Delivery
18. © 2016 SAP SE or an SAP affiliate company. All rights reserved. Thank You! Doug Freud Associate Vice President Doug.Freud@sap.com