Advanced Analytics for the Non-Data Scientist:

What You Really Need to Know

Ashish C. Morzaria
SAP
Who Is This Webinar For?

Do you know what these terms mean?

 Triple Exponential Smoothing
 Simple Linear Regression
 K-Means Analysis

This session is really focused on:

 Business analysts, data analysts, IT people … Basically everybody but the Data Scientists! 
 Understanding the basic concepts to demystify “Advanced Analytics”
 Bringing advanced capabilities “to the rest of us”

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 1

Why an “Advanced Analytics” Session for “BI People”?

The problem is most BI customers don’t focus on “Advanced Analytics”

 “We are still dealing with the basics of BI first …” (or upgrading to BI 4.1)
 “We need to get a handle on our existing data before we create more!”
 “We don’t have any ‘Data Scientists’ and our BI people can’t do it.”

Only 21% of organizations have “fully” or “partially”

implemented predictive analytics.

61% are “still exploring” or “have no plan”!

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 2

 The Big Data Problem

Web, mobile, social & machine Faster Decision Cycles

generated data explosion

Advanced Analytics Skills Gap

“Demand for deep analytical talent in the United States could be 50 to 60% greater
than its projected supply by 2018.”
 The Solution: Automated Algorithmic Analysis

Descriptive BI Diagnostic BI Predictive BI Prescriptive BI

Describe Understand Predict Recommend

A human using
a computer

A computer
providing info
to a human
“Advanced Analytics” Is the Next Step in Business Intelligence

Where Is Your Organization on the Spectrum?

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 5

Spectrum of Users for Analytics

Data Data Executives/

Scientist Analysts Business Users
 Create complex predictive models  Transform and enrich data  Interact with published predictive
and simulations source(s) analysis
 Validate predictive business  Create simple predictive models  Visualize results in context of
requirements and simulations use case
 Publish results back to source  Visualize results and publish to BI  Collaborate with colleagues toward
Platform closure/action

.1% ~3% 97%

Representative
User Base

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 6

Today, Predictive Analytics Is an Island

Specialization
Chasm
PM
LOB DM
ETL
PA

ETL

Reporting Discovery

Cleaned Data BI
EIM Analysis Dashboards
Raw Data

Sophistication / Skill Set

There is a disconnect between the BI and Advanced Analytics world

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 7

Bringing Predictive Analytics to Business Users Is Key

Specialization

PM
LOB DM
ETL Automated
Predictive PA

“Invisible
Analytics” in BI
Reporting Discovery
BI
Cleaned Data BI
EIM Analysis Dashboards
Raw Data

Sophistication / Skill Set

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 8

Predictive in SAP Lumira
Example of Invisible Analytics

© 2015 SAP AG. All rights reserved.

 Invisible Predictive Analytics:
Top Influencers in Lumira

Powered by
Predictive
Analytics

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 10

 Invisible Predictive Analytics:
Time Series Forecasting in Lumira

Powered by
Predictive Analytics

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 11

Predictive Analytics
Very Visible Results

© 2015 SAP AG. All rights reserved.

Predictive Analytics: Very Visible Results

Make more money:

 Walmart discovered prior to hurricanes, customers bought flashlights, batteries … and Pop-Tarts1
 Best Buy discovered 7% of its customers account for 43% of its sales1

Reduce costs:
 A major Canadian bank:
– Increased campaign response rates by 600%
– Cut acquisition cost by 50%
– Boosted ROI by 100%2
 A European telecom reduced customer churn from 20% to 5% using predictive analysis1
 Airlines better estimate the number of passengers who won’t show up for a flight2

Save Lives:
 Health care: finding emerging symptoms for Ebola before the pattern is obvious to the naked eye
 Route optimization to reduce response times: how much is 5 minutes worth to a dying patient?
1 The Economist, The Data Deluge, “Data, data everywhere”, February 27,2010, pages 3-5
2 Wayne W. Eckerson, Predictive Analytics: Extending the Value of Your Data Warehousing Investment, TDWI Best Practices Report, 2007, page 6

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 13

Advanced Analytics*:
An Introduction

* “Advanced Analytics” and “Predictive Analytics” will be used relatively interchangeably in this presentation

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 14

What Is Predictive Analysis?

High level:
 Using algorithmic analysis to recognize data relationships that influence likely outcomes and identify potential
risks and opportunities before they occur to make better decisions in the future.
– Approximate the relationship between variables and their outputs and represent it as an algorithm (set of rules)
– Use the algorithm against future data to predict the response with the least amount of error

In the BI context:
 PA is another toolset that helps us uncover patterns and relationships algorithmically that can be used against
future data sets instead of relying solely on visual representations that require the analyst to infer the future

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 15

How Does Predictive Analysis Work?

Trivial Example:
 Analysis of an entire year’s worth of store sales data reveals the following patterns:
– Soda is bought 40% of the time when potato chips are purchased
– Purchases of umbrellas increase 125% during the months of March and April
– When baby formula is purchased by a male: 54% of the time, so is at least one case of beer

As a store manager, what would you do with this data?

 Promote sales of soda pop in the same isle as potato chips (Cross sell)
 Increase stock of umbrellas in the spring (Forecasting)
 Put the beer closer to the baby isle (Trending)

+
© 2015 SAP SE or an SAP affiliate company. All rights reserved. 16
Predictive 101
(Advanced analytics for the rest of us)

© 2015 SAP AG. All rights reserved.

What’s a Predictive Model?

A predictive model algorithmically represents the desired target within the dataset. This model
can then be applied to other (or future) datasets to identify elements that should be targeted.

Predictive Models:
 “This last transaction could be fraudulent”

Descriptive Models:
 “Male purchasers of baby formula are 56% likely to also buy a case of beer, but female purchasers are likely to
have a higher number of items and a larger total bill”

Decision Models:
 “Your credit score is pretty low and you have few assets, giving you a mortgage would be pretty risky …”

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 18

What Is the Predictive Process?

How much time is typically spent in each area?

 Business understanding: 5–15 percent

 Data understanding: 5–10 percent
 Data preparation: 50–60 percent
 Modeling: 5–15 percent
 Evaluation: 5–10 percent
 Deployment: 10–15 percent

The secret: Good analytics require good data.

Any problems in the data can directly affect the

(accuracy of) predicted outcomes.

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 19

How Are Models Created?

A model can be created by a Data Scientist:

 Requires strong understanding of the data as every observation/assumption defines the model
 Resulting model can be validated by applying to historical data to determine “Predictive Power”

SAP Predictive Analytics Automated Algorithms are adaptive and are self-training
 Model can be validated by applying against the “testing” set of historical data to determine “Predictive Power”
 System iterates continuously until predictive power is high enough
 Constant validation to determine when model needs to be modified (Model Manager)

Data Scientists can typically create “better” models:

 A trained Data Scientist who has spent days or weeks creating and validating models can have a more logical
model because it is based on semantic understanding of the data
 A programmatically created model relies on algorithmic pattern recognition, which typically cannot have
semantics added afterwards
 BUT: It takes them *much* longer

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 20

Where Do We Get The Information For Predictive?

Information about who they are:

Profile Friends / Comments / Interaction Comments

information Location from calls
Associations Feedback history

Information about their behavior:

What they What they put What they When they How they How much What else they What do they
searched for in their cart actually bought bought bought they paid bought buy regularly?

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 21

 SAP Predictive Analytics is available for 30-day Trial!
Where to Start? www.sap.com/trypredictive

What questions are you trying to answer for your company?

 “Where do my customers live, and what do they do for a living?”
 “How much will my wealthiest customers spend for my product or service?”
 “If my customers buy product A, what percentage will also buy product B?”
 “If I offer an online customer a discount for a product related to their shopping cart, will they buy it?”

Understand your data before you apply any algorithms!

 It is easier to ask questions if you know what type of data you have

Even seemingly simple questions have value:

 Your questions will get more complex and detailed over time, but don’t underestimate “Yes/No” results
 What if you could increase your customer response rate by even 20%? Would that be worth it?

Good News! There’s no “100% right” answer – There’s just different levels of “right”…

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 22

How Does Predictive Integrate In The Real World
Example Use Cases

© 2015 SAP AG. All rights reserved.

Name Gender Age Marital Recent Activity C-Sat Renewed
Before
Predicted
Churn Batch Scoring

NEW Data
(Current Customers)

Significantly increase ROI through dataset reduction:

Hancock, John M 38 D Y 4.2 N ? • Lower campaign costs by targeting those most likely to leave
Doe, Jane F 45 M Y 9.4 N ?
• Increase response rate by targeting even more specifically on other attributes
Red, Simply F 18 S N 2.1 N ?
• Increase C-Sat by not hassling loyal customers

Customer not expected to

Model churn, so don’t bother them!

Hancock, John M 38 D Y 4.2 N Y

Hancock, John M 38 D Y 4.2 N Y
Doe, Jane F 45 M Y 9.4 N N
Red, Simply F 18 S N 2.1 N Y
Red, Simply F 18 S N 2.1 N Y

Targeted List
Name Gender Age Marital Recent Activity C-Sat Renewed Before Churn
Real-Time Scoring

NEW Data
(Single Customer)

Real-time scoring enables automated decision-making:

Hancock, John M 38 D Y 4.2 N ?
• Make an offer while customer is still on the phone/web site
• Proactively address situations by recognizing patterns (i.e. shopping carts)
• Cross sell by offering items that are highly likely to be purchased

Model

Special Offer:
Hancock, John M 38 D Y 4.2 N Y
Renew now and save 25%!
 Embedding Predictive Analytics Into BI Workflows

Model Embedded as
Stored Procedure

Customer Database
Business Users can get on-the-fly
Hancock, John M 38 D Y 4.2 N Y scoring without even knowing they
SQL are using predictive algorithms
Doe, Jane F 45 M Y 9.4 N N

Red, Simply F 18 S N 2.1 N Y

Lumira Dataset w/ Scoring

Lumira Server for Teams

Lumira Storyboard Lumiar Server for BI Platform

SAP Lumira Cloud

 Predictive Algorithm
Cheat Sheet

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 27

Shhh! – The Secret to Advanced Analytics for BI People

You don’t need to know HOW the algorithms work, but you do need to know WHAT they are for …*

Advanced Analysis

Descriptive Algorithms Predictive Algorithms

Clustering Association Classification Time Series

… … … …

… … … …

… … … …

* Actually, it does help to know how they work, but it is not absolutely required to get started

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 28

 Predictive Offerings
From SAP

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 29

Market leadership

Victory Big Data Wisdom

Index Report Predictive Wave of Crowds

Hurwitz Forrester Howard Dresner

Victor Leader Top vendor
SAP offers a “fast time to “SAP is a leader due to a “The top vendors for
value and ability strong architecture and advanced and predictive
to support very large data strategy.” analytics include SAP.”
sets.”

30
SAP Solutions for the Entire Spectrum of Users

No Low High

Level of Skill Set – Analytics

Business Data
Business Users & LOB
Analysts Scientist

Embedded Analytics 97% Industry & Business

Process Analytics 3% Custom
Analytics >0.1%
SAP Lumira SAP PA: Automated Analytics SAP PA: Expert Analytics SAP PAL R Integration

SAP ADVANCED ANALYTICS

SAP Predictive Analytics: Expert Analytics
Self-Service for Business Analysts and Data Scientists

Provide Data Scientists and

Business Analysts with
sophisticated algorithms to take
the next step in understanding
their business and modeling
outcomes
 Perform statistical analysis on your data
to understand trends and detect outliers
in your business
 Build models and apply to scenarios to
forecast potential future outcomes
 Breadth of connectivity to access almost
any data
 Optimized for SAP HANA to support
huge data volumes and in-memory
processing

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 32

SAP Predictive Analytics: Automated Analytics
Data Scientist in a Box

Provide Business Analysts and Data

Scientists with a fully automated process
 Data preparation
 Create 1000s of derived attributes
 Define metadata once
 Builds analytic dataset automatically

 Predictive modeling/Data mining

 Regression/Classification
 Segmentation
 Forecasting
 Association rules
 Social Network Analysis

 Advanced model deployment and management

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 33

Predictive Analytics Solutions from SAP

Any Data Source

Relational Databases Big Data Sources CRM/ERP Stores Applications Cloud Services

SAP Predictive Analytics 2.x

Data Preparation Visualization Automated Analysis SDK/API Model Management

Scoring Social Recommendation Expert Analysis Connectors

SAP HANA Predictive Applications

In-Memory Processing Engine
25+ Industries  11+ LoBs
R
Predictive Analysis Automated
R-Scripts
Library Predictive Library
Financial & Retail & Public Sector O&G,
Telecommunications
Insurance Consumer & Healthcare Manufacturing
Services Products & Utilities

Transactions

Cloud / On-Premise

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 34

Wrap-Up
Closing Points

© 2015 SAP AG. All rights reserved.

 Key Messages

 SAP Predictive Analytics is designed to:

– Help companies make better use of their data – and better decisions
– Make more meaningful and proactive sense of new data coming in
– Answer difficult questions that BI is not equipped for

 Predictive Analytics is the next step in Business Intelligence:

– PA is a natural progression that customers don’t realize yet
– PA Return on Investment (ROI) – not TCO: Answer the ROI question first
– Almost every analytics use case could benefit from predictive algorithms

 Increasingly, Big Data = Predictive Analytics

– Algorithms are the key to finding needles in the Big Data (Hay)stack
– Automated Analytics enables more people to work with Big Data without data science experience

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 36

Come To The In-Person Predictive Sessions at ASUG BOUC!

2706: Roadmap: Predictive Analytics: What's New and What's Next

2765: Beginner's Guide to Harnessing Big Data and Internet of Things For Real-World Use Cases
3526: ASUG Predictive Analytics Influence Council
3627: SAP Predictive Analytics Solution – Hands-On Session
2762: Predictive Analytics For The Non-Data Scientist/BI User: What You Really Need To Know
2703: Top 10 Predictive Use Cases and Customer Case Studies
2513: Predictive Maintenance & Service: Customer Lessons Learned

ASUG SAP ANALYTICS & BUSINESSOBJECTS CONFERENCE

August 31 – September 2
Austin, Texas

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 37

Where to Find More Information

Predictive on SCN: http://scn.sap.com/community/predictive-analysis

Predictive Official Product Tutorials: http://scn.sap.com/docs/DOC-32651
Predictive Blog: http://scn.sap.com/community/predictive-analysis/blog
Predictive 30-day Trial http://www.sap.com/trypredictive
Predictive Product Roadmap * http://service.sap.com/roadmaps (then Analytics > Predictive)

* Requires login credentials to the SAP Service Marketplace

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 38

Thank you!

Ashish C. Morzaria, SAP

Director – Advanced Analytics
a.morzaria@sap.com

© 2015 SAP SE or an SAP affiliate company. All rights reserved.

© 2015 SAP SE or an SAP affiliate company. All rights reserved.

No part of this publication may be reproduced or transmitted in any form or for any purpose without the express permission of SAP SE or an
SAP affiliate company.

SAP and other SAP products and services mentioned herein as well as their respective logos are trademarks or registered trademarks of SAP SE
(or an SAP affiliate company) in Germany and other countries. Please see http://global12.sap.com/corporate-en/legal/copyright/index.epx for additional trademark
information and notices.

Some software products marketed by SAP SE and its distributors contain proprietary software components of other software vendors.

National product specifications may vary.

These materials are provided by SAP SE or an SAP affiliate company for informational purposes only, without representation or warranty of any kind, and SAP SE or its
affiliated companies shall not be liable for errors or omissions with respect to the materials. The only warranties for SAP SE or
SAP affiliate company products and services are those that are set forth in the express warranty statements accompanying such products and services, if any. Nothing
herein should be construed as constituting an additional warranty.

In particular, SAP SE or its affiliated companies have no obligation to pursue any course of business outlined in this document or any related presentation, or to develop or
release any functionality mentioned therein. This document, or any related presentation, and SAP SE’s or its affiliated companies’ strategy and possible future
developments, products, and/or platform directions and functionality are all subject to change and may be changed by SAP SE or its affiliated companies at any time for
any reason without notice. The information in this document is not a commitment, promise, or legal obligation to deliver any material, code, or functionality. All forward-
looking statements are subject to various risks and uncertainties that could cause actual results to differ materially from expectations. Readers are cautioned not to place
undue reliance on these forward-looking statements, which speak only as of their dates, and they should not be relied upon in making purchasing decisions.

© 2015 SAP SE or an SAP affiliate company. All rights reserved. 40