Copyright © 2013 by KNIME.com AG. All rights reserved.
For questions and suggestions please contact [email protected]
Leveraging SAS with KNIME
Webinar
Phil Winters
with
Dr. Rosaria Silipo
Copyright © 2013 by KNIME.com AG. All rights reserved. KNIME and R: the best of two worlds 2
Welcome!
• Thank you for coming!
• You are muted, so questions to KNIME Host by chat
• Links to the slides and workflows by email later.
Copyright © 2013 by KNIME.com AG. All rights reserved.
SAS and all other SAS Institute Inc. product or
service names are registered trademarks or
trademarks of SAS Institute Inc. in the USA and
other countries. ® indicates USA registration.
Copyright © 2013 by KNIME.com AG. All rights reserved.
Course Material
• Webcast
• Data
• Example workflow
• Slides of this presentation
KNIME Server Training 4
Copyright © 2013 by KNIME.com AG. All rights reserved.
Goals of Today’s Course
KNIME User Training 5
• Why a course to leverage SAS with KNIME?
• Intro for SAS Users: KNIME
• An Example: Next Best Offer
• Package, Reusing and Surfacing to Business Users
• Where to find more examples and help
Copyright © 2013 by KNIME.com AG. All rights reserved.
Access Manage
AnalysePresent
A bit of History:
The Original SAS Concept
Copyright © 2013 by KNIME.com AG. All rights reserved.
Access Manage
AnalysePresent
More Products
to meet new requirements
DB
2
DI
RT
D
Macro
PC
FIL
IE
Ora
cle
EG
DVD
SCL
Connect
JMP®
ET
S
OR
AF
Gra
ph
IDP
STA
T
QC
IML
EM
MM
Copyright © 2013 by KNIME.com AG. All rights reserved.
Access Manage
AnalysePresent
Even More Products….
DB
2
DI
RT
D
Macro
PC
FIL
IE
Ora
cle
EG
DVD
SCL
Connect
JMP®
ET
S
OR
AF
Gra
ph
IDP
STA
T
QC
IML
EM
MM
HadoopTeradata
PM
ML …
Model M
anager
R…
. in
IM
L
Text Mining
Hig
h P
erf
orm
ance A
naly
tics
Social Media Analytics
Copyright © 2013 by KNIME.com AG. All rights reserved.
Our new reality:
Choice and Control
New Data
Other
Applications
New
Business
Challenges
New
Methods
New
Science
New
Infrastructures
Copyright © 2013 by KNIME.com AG. All rights reserved.
The KNIME Platform
Open, Open Source, Free on the Desktop
10
Copyright © 2013 by KNIME.com AG. All rights reserved. 11
Copyright © 2013 by KNIME.com AG. All rights reserved.
The question is not “which is better”
The question is: What’s the Big Difference?SASA script-oriented 4GL programming language in
four major parts:
• The DATA step
• Procedure steps
• A macro language,
a metaprogramming language
• ODS statements
• GUIs: are most often merely front-
ends to facilitate SAS Program script
generationWiki Descriptions….
KNIME User Training 12
KNIMEA Script-free environment
comprised of
• Nodes and Connectors
• Metanodes, Flow variables and
Connectors for packaging
• GUI is the interface
Copyright © 2013 by KNIME.com AG. All rights reserved.
Statistics
Data Mining
Machine Learning
Web Analytics
Text Mining
Network Analysis
Social Media Analysis
WEKA
R
Community / 3rd
MySQL, Oracle, etc.
SAS, SPSS, etc.
Excel, Flat, etc.
Hive etc.
XML, PMML
Text, Doc, Image
Web Crawlers
Industry Specific
Community / 3rd
ETL
Row,
Column
Matrix
Text, Image
Time Series
Java
Python
Community / 3rd
KNIME: Free on the desktop including
over 1000 native and embedded nodes:
R
JFreeChart
Community / 3rd
via BIRT
PMML
XML
Databases
Excel, Flat, etc.
Hive etc.
Text, Doc, Image
Industry Specific
Community / 3rd
13Some 3rd party nodes are charged
Copyright © 2013 by KNIME.com AG. All rights reserved.
An Example: “Next Best Offer!
• Accessing various data sources
including social media analysis
• Consolidate and Transform the data
• Explore and Visualize the data
• Apply various predictive methods to the data
• From various packages!
• Automatically determine the best model
• Deploy !
KNIME User Training 14
Copyright © 2013 by KNIME.com AG. All rights reserved.
Accessing Data
KNIME User Training 15
Huge
Range of
Read
(and write)
nodes
available
At no cost.
Each Node
has a
specific
dialog
SAS Data
can be read
with one
click
Tables
available at
every step
Copyright © 2013 by KNIME.com AG. All rights reserved.
Transforming data
KNIME User Training 16
Copyright © 2013 by KNIME.com AG. All rights reserved.
Packaging Workflows
KNIME User Training 17
Drage and Mark
nodes,
Right Click,
Choose Colapse..
To create a
Metanode
that can be reused
Copyright © 2013 by KNIME.com AG. All rights reserved.
Our Example: “Next Best Offer!
KNIME User Training 18
The Access and
Transform
Metanode we
created
Copyright © 2013 by KNIME.com AG. All rights reserved.
More Transformation
KNIME User Training 19
KNIME nodes
for transforming
(alternate to
formatting)
Calling your Favorite External
Packages such as Java, Python,
Matlab, SQL, REST,
Or your favorite Database
Applying Color
to Attributes
Copyright © 2013 by KNIME.com AG. All rights reserved.
Using SAS Transformation
KNIME User Training 20
Calling SAS either
Locally or Remotely.
Automatically passing
Data into and out of
SAS
Controlling how that
happens with Flow
Variables and
Quickforms
Copyright © 2013 by KNIME.com AG. All rights reserved.
Giving you instant Customized Dialogs
KNIME User Training 21
No Macro Code. No
SCL. No Java. No
Other Programming
Language required.
A Nice Dialog for many
types of users…..
Copyright © 2013 by KNIME.com AG. All rights reserved.
Explore and Visualize
KNIME User Training 22
Many nodes
For Exploring
and
Visualizing
All that allow
Marking and
highlighting
across all
nodes
No seperate package
required.
Other packages can be
used if available
Copyright © 2013 by KNIME.com AG. All rights reserved.
Be Open to All Methods,
and let the best model win !!!
KNIME Decision Tree
KNIME Logistic Regression
R Decision Tree
PMML model (from SPSS)
Other Methods
Weka
Dymatrix Uplift
SASKNIME User Training 23
Copyright © 2013 by KNIME.com AG. All rights reserved.
Allow for best practices without
learning scripting
KNIME User Training 24
Missing Management
Partitioning
Binning and Bagging
Boosting
Logic and Flow Control
Cross Validation
Feature Elimination
Feature Selection
Error Handling
Copyright © 2013 by KNIME.com AG. All rights reserved.
Report with your favorite Tool
KNIME User Training 25
Or use the
built-in and Free
BIRT open source BI Tool
Copyright © 2013 by KNIME.com AG. All rights reserved.
Deploy the Models
KNIME User Training 26
Score any Database
(including Hadoop)
Use the PMML on another
system (Such as ADAPA)
Or how about scoring
straight into SAS via SQL
and the (commercial
node!) from Dymatrix?
Copyright © 2013 by KNIME.com AG. All rights reserved.
Deploy to Excel…..
Or directly back to SAS!
KNIME User Training 27
SAS Transport Dataset
Copyright © 2013 by KNIME.com AG. All rights reserved.
Deploy: Types of Users
28
• Wrap new nodes
• Wrap legacy software
Developers
• Develop workflows
• Provide templates
Power Users
• Modify templates
• Run parameterized
workflows
Business
Users
• Launch workflows
• Create reports
Business
Consumers
• Consume Reports
• Exception
Notification
Management
Copyright © 2013 by KNIME.com AG. All rights reserved.
Power Users in Teams The KNIME Server (Commercial Software)
29
Copyright © 2013 by KNIME.com AG. All rights reserved.
Embedding, Automating, SecurityThe KNIME Server (Commercial Software)
30
KNIME Server
KNIME
Copyright © 2013 by KNIME.com AG. All rights reserved.
Business Consumers via the webThe KNIME Server (Commercial Software)
31
Copyright © 2013 by KNIME.com AG. All rights reserved.
When to use KNIME?
New Data Sources
New Methods Required
New Approaches
New User Groups
Hot Topics
New Money required for Extensions ????????????????
KNIME User Training 32
Copyright © 2013 by KNIME.com AG. All rights reserved.
Hot Topics with KNIME (and SAS!)
users:
Social Media Analysis
Recommendation/Next Best Offer/Market Basket/etc. Analysis
Text Mining
Network Analysis
Realtime
Machine Learning
Making R Usable
Big Data
Data Mining Automation with DYMATRIX DynaMine
Campaign Optimization with DYMATRIX DynaCampaign33
Copyright © 2013 by KNIME.com AG. All rights reserved.
Hot Topics: Customer Risk
34
Copyright © 2013 by KNIME.com AG. All rights reserved.
Hot Topics: Social Media Analysis
35
Copyright © 2013 by KNIME.com AG. All rights reserved.
Hot Topics: Text Mining
36
Copyright © 2013 by KNIME.com AG. All rights reserved.
Hot Topics: Network Analysis
37
Copyright © 2013 by KNIME.com AG. All rights reserved.
Hot Topics: Text + Network Mining
38
Copyright © 2013 by KNIME.com AG. All rights reserved.
Hot Topics: Recommendation Engine
39
Copyright © 2013 by KNIME.com AG. All rights reserved.
Hot Topics: Realtime Scoring
40
Copyright © 2013 by KNIME.com AG. All rights reserved.
Hot Topics: Big Data
41
Logfiles.gz
Define Hive Tables / Views
Hadoop Cluster
KNIME: Access Big Data with Database Reader
Copyright © 2013 by KNIME.com AG. All rights reserved.
How to start with KNIME?
• Download and install KNIME
• Download and install the needed
(FREE!) Extensions
• Read the Intro to SAS Users Book
(FREE!)
• Download and install this example
(FREE!)
• Download and install other examples
(FREE!)
• Enjoy being FREE to choose…
KNIME User Training 42
Copyright © 2013 by KNIME.com AG. All rights reserved.
For this example to work:
Install KNIME
Install R Extension (via KNIME Pulldown)
Install SAS7Bdat Extension (via KNIME Pulldown)
Install extra R Extension SASxport.
Import Workflow
Place all data in c:\data
Pay For SAS (if you want to use it with KNIME).
KNIME User Training 43
Copyright © 2013 by KNIME.com AG. All rights reserved.
Further Resources
• KNIME for SAS Users (weblink…)
• KNIME pages (www.knime.org)• APPLICATIONS for example workflows
• RESOURCES with links to a number of resource pages, like
downloads, updates, documentation, ...
• KNIME Tech pages (http://tech.knime.org/)• FORUM for questions and answers
• DOCUMENTATION for documentation, FAQ, changelogs, ...
• LABS where to find new experimental nodes
• COMMUNITY CONTRIBUTIONS for development instructions and third
party nodes
• KNIME TV channel on
KNIME Server Training 44
Copyright © 2013 by KNIME.com AG. All rights reserved.
The KNIME Platform
Open, Open Source, Free on the Desktop
45
Copyright © 2013 by KNIME.com AG. All rights reserved. Copyright © 2013 by KNIME.com AG. All rights reserved.
Thank you