Abstract: SAP Predictive Analytics 2.3 bridges the gaps between IT, Business and Analytics. Learn how SAP Predictive Analytics 2.3 leverages current SAP infrastructure and provides the data preparation, visualization and data modeling tools necessary to gain insights from your data.
Join us as we will take a look at SAP Predictive Analytics (PA) 2.3 and demonstrate the value by analyzing data that will help drive real-world decisions:
• Understand where SAP PA 2.3 resides in SAP’s analytics roadmap
• Outline system and hardware requirements for implementation
• Learn how SAP PA 2.3 is not just for Data Scientists
• Demonstrate capabilities within SAP PA 2.3
• Answer questions and discuss how PA 2.3 can improve your business
SAP Applications and the Modern Data Scientist - Predictive Analytics for the End User
1. SAP Applications and the Modern Data Scientist –
Predictive Analytics for the End User
2. Introductions
What is Predictive Analytics
SAP Predictive Analytics 2.3 Overview
Where SAP is in the Advanced Analytics Market
System Demonstration
Use Case: Association Analysis
Use Case: Regression
Questions/Next Steps
3. 3
On the Phone:
Rob Jerome
Vice President, Innovation + Technology
rob.j@dickinson-assoc.com
Todd Siedlecki
Consultant, Predictive Analytics Practice Lead
todd.s@dickinson-assoc.com
Olavo Figueiredo
Consultant
olavo.f@dickinson-assoc.com
4. 4
We Are:
Focus: Delivery of quality SAP Business Suite, BI/Analytics,
and Mobility consulting services to customers across
North America, Europe, and Asia.
Our People: A team of 140+ full-time SAP professionals reflects the
ideal mix of years of relevant business knowledge, very
strong SAP credentials, and solid communication skills.
Our team has an average of 15 years SAP and 19 years
business experience.
Offices: Chicago, IL (Headquarters)
Satellites: New York, NY | Scottsdale, AZ | Cincinnati, OH
We are:
5. 5
Experience
What Sets Us Apart? Our People.
Experienced consultants with strong SAP knowledge, sound project
management capability, and years of industry experience.
Proven experience in delivering innovative ERP solutions with
minimal disruption to the business.
An open corporate culture that makes us
“big enough to deliver value and small enough to care”.
We carefully create each project team or support team
to match the client objectives and its culture.
Most important, we understand and believe strongly that
Companies don’t implement SAP… People Do.
No.TeamMembers
0 – 3yrs 3 – 8yrs 8 – 14yrs 14+ yrs
6. 6
Partnership and Designations
SAP Gold Channel Partner
SAP Services Partner
SAP All-in-One Certified Solutions
SAP-Qualified Partner for RDS
Business Objects
Sybase Partner
SuccessFactors Partner
10. 10
SAP Predictive Analytics - Myths
Requires a Ph.D. to implement
Hard to execute without technical
expertise
Does not require business input
Only for large companies
20. 20
Predictive Analytics Process
Model deployment,
scoring, monitoring
Define the objectives of
the analysis;
Understanding the
business problem
Data selection,
cleansing,
transformation; initial
data exploration
Model building, training,
testing, evaluation
Reiterate
21. 21
Classes of Applications
Time Series Analysis
Classification Analysis
Cluster Analysis
Association Analysis
Outlier Analysis
22. 22
Time Series Analysis
Use past data points as the basis for projecting future
values
Variable = Data (i.e. Sales or Headcount) with a series of
values over time
Historical patterns of past data are used to make
predictions
23. 23
Classification Analysis
Goal is to predict a variable (a.k.a. target or dependent
variable) using the data of other variables
Largest group of applications of predictive analysis
Examples: churn analysis, target marketing, predictive
maintenance
24. 24
Cluster Analysis
Takes the data set and groups it into segments (clusters)
that have similar attributes
Application is often used to subset a large data set in
order to better understand the attributes of the smaller
subsets
Helps to find patterns and explanations for relationships
Examples: customer segmentation
26. 26
Outlier Analysis
This class of applications seeks unusual or unexpected
values in the dataset
Possible significant impact on predictive models, so it’s
used in the context of all other classes of predictive
applications
Could be genuine variations or errors
Example: fraud detection
28. 28
SAP Predictive Analytics 2.3 - Overview
Automate data prep, predictive modeling, and
deployment – and easily retain models
Harness in-database predictive scoring for a wide variety
of target systems
Leverage advanced visualization capabilities to quickly
reveal insights
Integrate with R to a enable a large number of algorithms
and custom R scripts
Deploy SAP Predictive Analytics stand-alone or with
SAP HANA
29. 29
SAP Predictive Analytics 2.3 – System Requirements
Server Requirements
300 MB of disk space
2GB of RAM
Client Hardware Requirements
150 MB of disk space
512 MG of RAM
30 day free trial available
http://go.sap.com/product/analytics/predictive-analytics.html
30. 30
SAP Predictive Analytics 2.3 – Automated vs. Expert
Automated Analytics
Designed for business
analyst or super user
Drag and drop/Point and
click tool
Preps data for the user
Automatically selects
appropriate model
Expert Analytics
Designed for statisticians
Robust functionality with
statistical software R
Create your own algorithms
Compare effectiveness of
models
32. 32
Demo 1 – Business Problem
Background
A manufacturing company is seeking to lower their
preventative maintenance costs on certain machines
33. 33
Demo 1 – Predictive Maintenance
Maintenance scheduled
according to set time period
Future State – Predictive
Maintenance
Maintenance scheduled
according to data analysis
Current State – Preventative
Maintenance
35. 35
Demo 2 – Business Problem
A marketing company is experiencing a high rate of
turnover among employees
When an employee leaves, the process is very
expensive due to the following:
Lost Knowledge
Training Costs
Interviewing Costs
Lowered Productivity
36. 36
Demo 2 – Analytics to Improve HR
HR would like to use analytics to know not only which
employees will be likely to leave, but also take a more
refined approach by grouping employees with similar
characteristics together
Goals:
Segment out employees into different groups
Determine which groups are most likely to have a high turnover rate
Analyze data to determine what incentives could be best offered to
keep employees from leaving
38. 38
What’s Next?
Q+A
Contact Todd Siedlecki to discuss how SAP
Predictive Analytics may fit in to your analytics
strategy
Email – todd.s@dickinson-assoc.com