Moderator: - PowerPoint PPT Presentation

1 / 30
About This Presentation
Title:

Moderator:

Description:

Two-Step Cluster Visualization SDK. Scoring Engine Na ve Bayes. Web Importer Neural Network ... Two-Step Cluster. Sequence Association. C&RT. CARMA. Tree Viewer ... – PowerPoint PPT presentation

Number of Views:76
Avg rating:3.0/5.0
Slides: 31
Provided by: paulas88
Category:

less

Transcript and Presenter's Notes

Title: Moderator:


1
  • Moderator
  • Brian Sterrett
  • VP and General Manager, Enabling Technologies
    Division, SPSS Inc.

2
Agenda
  • Predictive Analytics Market Opportunity
  • SPSS Enabling Technologies
  • Case Study Clementine 8.5
  • Case Study Alterian Modeling 1.0
  • QA

3
Who is SPSS?
  • Leadership
  • Market leader in Predictive Analytics
  • Experience
  • 250,000 customers
  • 30 year heritage in analytic technologies
  • Stability
  • Founded in 1968
  • 210M in revenue FY2003
  • NASDAQ SPSS
  • Analytics gold standard
  • 80 of Fortune 500 are SPSS customers

4
Predictive Analytics Defined
  • Predictive analysis helps connect data to
    effective action by drawing reliable conclusions
    about current conditions and future events.
  • Gareth Herschel, Research Director, Gartner
    Group

5
Documented ROI for Predictive Analytics
Investment and overall return predictive
projects versus nonpredictive projects
Source IDC, 2003
6
Proven Success ROI
Industry
Company
Need
Results
Reduced manualreviews by 40
Fraud Detection
Retail
Web Analytics
9M return in 5 months
Entertainment
7
The Analytic Enterprise
8
Sample Applications Powered by SPSS
  • Predictive CRM
  • Real-time offer optimization
  • Attrition / churn management
  • Targeted marketing / segmentation
  • Fraud detection and analysis
  • AML
  • Telco
  • Healthcare billing
  • Demand Forecasting
  • Risk Management

9
SPSS Enabling Technologies Division
  • We help software companies grow their business
    with predictive analytics applications Powered by
    SPSS.
  • SPSS delivers the OEM technology, expertise and
    brand recognition that enables our partners to
    succeed.

10
World-class OEM customers
11
Were OEM Ready
  • Expertise
  • Modeling
  • Predictive Analytics Application Design
  • Applications
  • PredictiveMarketing
  • Tools
  • SPSS Lexiquest
  • Clementine
  • Components
  • Time-Series Concept Extractor
  • Two-Step Cluster Visualization SDK
  • Scoring Engine Naïve Bayes
  • Web Importer Neural Network

12
Clementine Utilization of Enabling Technologies
  • Laura Squier
  • Sr. Product Manager, Clementine, SPSS Inc.

13
The Clementine Data Mining Workbench
  • World-class solution for highly scalable data
    mining and rapid deployment
  • Supports the entire data mining process
    including
  • Data access
  • Data understanding and visualization
  • Data preparation
  • Modeling using
  • Machine learning technologies
  • Statistical analysis techniques
  • Model understanding
  • Model deployment

http//www.spss.com/spssbi/clementine/index.htm
14
Advantages of Leveraging SPSS Components
  • Provide major product release every year
  • Minimize development time
  • Re-use rather than, rather than re-build
    capabilities
  • Use development resources who have domain
    expertise for the particular capability
  • The statistics team provides statistical
    components
  • The visualization team provides the visual
    components
  • Quickly implement
  • Components are delivered with well-designed
    programming interfaces facilitating easy
    integration into the Clementine code base
  • Deliver more features in less time

15
Component Utilization in Clementine Technologies
  • SPSS nViZn technology - development platform
    offering advanced visualization techniques,
    analytics, and interactivity
  • Multinomial Logistic Regression
  • OLS Regression
  • Factor and PCA Analysis
  • Two-Step Cluster
  • Sequence Association
  • CRT
  • CARMA
  • Tree Viewer
  • Cluster Viewer
  • SmartScore
  • Scoring Engine
  • Data understanding and visualization
  • Modeling using
  • Statistical analysis technologies
  • Machine learning
  • technologies
  • Model understanding
  • Data scoring and model deployment with Cleo and
    Solution Publisher

16
Implementation of Modeling Components
  • In Clementine, an input adapter gets the data
    from the Clementine Stream and sends it to the
    component.
  • The component processes the data, creates the
    model, and produces it with other information as
    output.
  • Clementine retrieves the output.

17
Results
  • Shorter development cycles
  • Fully supported and QAed components, ready to
    implement
  • More than doubled the features added to
    Clementine in one release.
  • More productive and robust technology that
    attracts new customers and retains existing
    customers.

18
  • Case Study
  • Alterian Embeds
  • Best-of-Breed
  • Data Mining Components
  • Richard Clarke
  • Product Manager, Alterian

19
Alterian
  • High speed Analytical Database Platform
  • Initially
  • Enabling technology
  • Partners producing applications
  • More recently
  • Applications
  • Marketing focus

20
The Requirement
  • Gap in the data mining area
  • In particular for predictive modeling and
    clustering
  • For applications such as
  • Response analysis
  • Churn analysis
  • Customer segmentation
  • Needed to address quickly

21
Options
  • Create
  • Too time-consuming and not enough internal
    expertise in this area
  • Acquire
  • Not much choice and expensive unless choose an
    unknown
  • OEM
  • Established technology at a reasonable price

22
Why SPSS?
  • Trusted brand
  • Breadth of coverage
  • Componentized algorithms

23
The Components
  • Linear regression
  • Future profitability or spend prediction
  • Logistic regression
  • Propensity to respond/churn prediction
  • Clustering
  • Identification of groups of similar behaving
    customers
  • CandRT
  • Rule detection in relation to specific behaviours
  • Associations
  • Basket analysis
  • Sequential patterns
  • Identification of related customer actions

24
The Integration
  • Picked simplest first linear regression
  • Well documented API
  • SPSS experts on hand
  • Standard PMML output model
  • Weeks to completion

25
The Integration
  • Commonalities across components
  • Made subsequent components integration easier
  • Ability to expose subsets of options
  • Integration of 5 out of 6 components in 3 months
  • Launched in April 2003

26
Alterian Modeling
  • A broad data mining application
  • Running on Alterians rapid Engine database
  • Totally integrated with other Alterian
    applications
  • A suite that provides end-to-end data mining
    capability through
  • Data exploration
  • Data preparation
  • Model building
  • Model validation and interpretation
  • Database scoring
  • Actioning, via campaign management

27
Market Impact
  • Can now offer a complete Marketing Suite
  • Has been a great up-sell opportunity
  • Two clients bought the same day they saw it
  • Has also leveraged full suite sales
  • 50 organizations now using Alterian Modeling

28
Where Next?
  • Expose more options within components (for expert
    usage)
  • Finish integrating sequential patterns component
  • Automate more of the data mining methodology
  • Integrate more SPSS components, such as
  • Tree Viewer
  • Cluster Viewer
  • Scoring Engine

29
QA
30
Thank You
  • Contact Information
  • Brian Sterrett
  • (312)651-3658
  • Bsterrett_at_spss.com
  • Log onto our website for more information
    www.spss.com/spssetd
  • or call 800-259-1028
Write a Comment
User Comments (0)
About PowerShow.com