TIBCO Spotfire: Data Science in the Enterprise

From Data to Insights
inInternet Time
DataScienceintheEnterprise
Eric Novik
IndustryAnalyticsGroup
TIBCOSpotfire

21st CENTURY PLATFORM
RV, EMS, FTL
ActiveMatrix
BPM
MDM
BusinessEvents
ActiveSpaces
LogLogic
In-Memory
Spotfire
Visual Analytics
Data Discovery
Statistics
Silver Fabric
Private Cloud
Public Cloud
SAAS Apps
tibbr
Social Media
Collaboration

© Copyright 2000-2013 TIBCO Software Inc. All rights reserved. TIBCO Confidential & Proprietary Information.
Are Your Models Behaving Badly?

Data
Mashups
Exploratory
Data Analysis
Features
Spotfire
Dashboards
Model &
Predict
Deploy
Champion
Model
Test &
Learn
Sales
Loyalty
Social
Media
Behavior
Filter
Map
Merge
Shape
Social
Network
Members
ImproveGuided -------- Deploy -------- In-LineExplore Data
Aggregate
Prepare DataBusiness Case
Increase
Productivity
Grow
Revenue
Ensemble
Forest
Regression
Additive
Models
Transaction
Visualize
Geo
Location
Service
Challenger
Models
At Rest
In Motion
Value
Theses
Reduce
Risk
ROI
Transaction
Dashboard
Updates
Data Science 2.0: Guided and In-Line Analytics

Time
The TIBCO Spotfire Experience Reduces Latency
Business
Value
Business Event
Data Ready for Analysis
Analysis Completed
Decision Made
$$$$
$$$
$$
$ Action Taken
TIBCO Spotfire
Analytics
Speeds Action
and Increases
Business Value

Data a Information a Knowledge a Wisdom
Single BI Platform
Traditional BI and Discovery
Design
Data Models
Source The Data
Build Reports
Business UsersJoe Data-Modeler
CRM ERP Financials
ETL
OLAP
Cube
Requests
Data Warehouse
Iterate the Loop

Business AnalystsJoe Data-Modeler
OLAP
Cube
BI Platform
Guided, Self-Service Analytics
Design
Data Models
Source The Data
CRM ERP Financials
ETL
Data Warehouse
The Spotfire Experience: Guided, Self-Service Analytics
Spotfire 5.x
In-Database Analytics
In-Memory Analytics
Hybrid On-demand Analytics
Data Virtualization
TIBCO Enterprise Runtime for R

Dynamic Visual Analytics: Data and Models

My Kind of Pie Chart….

“Numerical quantities focus on expected values,
graphical summaries on unexpected values.”
- John Tukey, Exploratory Data Analysis, 1977.
Visual Analytics

Visualizing Risk

Visualizing Correlations and Linear Models

Customer Acquisition/Retention
► Understand transactions and attractive
products for individual investors
► Identify prospects through data-mashups
of in-house and industry sources
► Understand client dissatisfaction
#1. Business and Customer Insight
Customer Growth
► Identify persuadable customers for
cross-sell / up-sell programs
► Drive campaign efficiencies
► Discover sensitivities of acceptance to
demographics and behavior
► Test and learn marketing and loyalty
Business leaders and marketing groups need insight in to prospects and
customer behavior and to identify “persuadables” for growing the business

Customer Life Cycle
Cross-sell / Up-sell
Sentiment & Intent
Analytics
Sales Force Detailing
Optimization
Test & Learn
Market Mix Analytics
Price Elasticity
Campaign
Effectiveness
Attrition & Social
Churn
Customer
Acquisition
Customer
Retention
Relationship
Growth
Customer
Lifecycle

Customer Growth & Cross-Sell
• Response
• Acceptance of product (wealth
management services, credit card)
• Analysis Unit
• Customer
• Features
• Bank transactions: ATM, checking; deposit,
withdrawal, balance, …
• Demographics: sex, age, profession,
nationality
• Model
• Classification trees
• Association & machine learning
• Value
• Customers * Value * Conversion

Citibank saw a 21% increase in
customer spend, and shortened
time-to-market for new campaigns
by 92% after implementing TIBCO
Spotfire cross-sell programs
Citibank Asia
http://www.tibco.com/customers/#citibank-asia

Citibank COP: Delivered Business Functionality
COP
360 Degree
Customer View
DWH
ACWS
AFEWS
CBOL
ECCM
GRB
ECS
Oracle
Master
data
download
Master
data upload
Offer
palette
Service Request &
Offer Feedbacks
Addr
change
Status
Changes
Offer
response
Financial &
Non-financial
transaction
Campaign
Definition, Offer
Request &
Feedbacks
Customer Citibank Phone Officer (CPO)
DM Staff Members
CTI
Offer
palette
Offers
Files
Campaign
Master &
Offer List
Business Rules
Definitions
KPIs, Performance
Metrics, Campaign
Feedbacks
Campaign
Definitions &
Offer list
Status
Changes
Offer
Request &
Feedbacks
Offer Request &
Feedbacks
Offer
response
Offer
response
Offer
response
Offer
request
Login
Internet offers
Campaign
Definition,
Limited Master
data upload
FMS
Offer
Feedbacks
Campaign
Definition,
Offer
Feedback
Fulfillment
info
 Intelligent Customer/Scenario Dialogue
Real-time synchronization of customer contact / touch-points (inbound and outbound)
 Dynamic Offer Management
Real-time (re) scoring
Real-time offer eligibility and prioritization
 Conversion Tracking and Analysis
Real-time dash boarding / tracking
 Dynamic Channel Synchronization and Management
Infrastructure to address time-sensitive offers (e.g.. real-time cross-offers on SMS based
on card transactions at specific locations)
 Sales Process Tracking
Close-loop tracking of intermediate fulfillment processes
 On-Board Measures, Statistics and Visibility
Reliability, Availability, Scalability
Hot – Dynamic Configuration

BMO saw a 15% increase in
customer spend on other products
after implementing TIBCO Spotfire
cross-sell programs.
Bank of Montreal

BEDiscover pattern
in data
…
Build predictive
statistical model
Deposits
correlate with
opening wealth
account in
some
demographics
Predict new
wealth mgt
customer from
account
transfers and
demographics
Export optimal
rules/thresholds
and deploy with BE
Make predictions
and act on them
at every customer
interaction
Measure
campaign
success
Business Analyst Data Scientist
Manager
5-10% increase in
wealth account
conversion rate
Male 28-42;
median deposit; $
transfers;
then extend offer
Refine/Extend
Extend selective offers
to open new wealth
account based on real-
time predictions of
propensity to accept
offer at time of events
e.g. transfer/deposit
Analyst
Event Cloud
Teradata Oracle
BPM
Tibbr
Analytics-to-Action: Cross-Sell / Up-Sell
TERR

Customer Insight: Social Media & Sentiment Analytics
• Response
• Customer sentiment / intent
• Analysis Unit
• Tweets
• Features
• Favorability
• Recommendation
• Influencers / Followers
• Models
• tm, twitteR, wordcloud
• Value
• Attribution : contribution of channel
• Optimization : scenario planning
• Allocation : media mix

Sentiment and Social Media Analytics

Influencers and Followers

Business Insights: Procter and Gamble Business Sphere
Source: McKinsey Quarterly, Nov 2011

Business, Customer and Consumer Insights
• Analyses
• Business Insights
• Customer Insights
• Consumer Insights
• Analysis Unit
• Product * Market
• Customer behavior
• Features
• Product/channel attributes
• Promotions, pricing
• Value
• Focus on product * market combinations
• Identify managers / attributes with issues

Enterprise Risk Aggregation
► Aggregated view of multiple risk positions by
geography, product, …
► Assess loss given default for products and
counterparties
#2. Risk Management
Credit and Counterparty Risk
► Handle unexpected changes in markets with
rapid rebalancing
► Incorporate predictive analytics e.g. stress
tests, economic capital
► Collaborate between traders and managers
Operational Risk
► Track desk in near real time
► Identify violations, patterns, rule-brushing
traders
Risk managers need interactive and aggregate views on risk; and ability
to visualize complex risk analyses with management
Market Risk
► Mark to market, VaR, scenario and sensitivity
analysis
► Interest rate and FX-risk
► Collaborate across organization for rapid
response to market conditions

Royal Bank of Scotland

Equity Trading
► Analyze price / volume across holdings with
drill down into individual securities
► Calculate statistical and technical indicators,
benchmark comparisons
► Incorporate indicators and financial data e.g.
Bloomberg, Thomson, FactSet
► Collaborate across trading teams
#3. Trading Analytics
Fixed Income Analysis
► Identify unknowns: bond prices, interest,
durations, convexity and yield
► Compare fixed income instruments with
economic indicators
► Incorporate statistical methods applied to
new and existing instruments
Trade Fraud and Compliance
traders
Traders need rapid and comprehensive view of risk and sensitivities

Portfolio Optimization
► Analyze risk-return sensitivities
► Efficient frontier statistics
► Drill to efficient portfolios of interest,
incorporating predictive analytics
► Rebalance on –demand
#4. Portfolio Management
Performance and Attribution
► Dimension-free exploration of assets across
time
► Real-time analysis and review of fund
strategies impact on performance
► Rapid transaction reconciliation with
exceptions analysis
► Integrate with CRM data for customer insight
Managers require multiple assets and funds in single view; need to prove
value and comply with new regulations
Regulation Compliance
► Comply with new regulations that require
thorough understanding of trading across
assets and liabilities

Anti-Money Laundering
► Identify / trace unknown and suspect
transactions across context e.g. business lines
and geographies
► Set thresholds, define segments and automate
alert scoring
► Comply with OCC guidelines
#5. Fraud and Compliance
Credit Card Fraud
► Identify fraudulent transactions by vendor,
location
► Highlight patterns, context and identify root
cause
Trade Fraud and Compliance
traders
Regulators expect banks to have powerful and robust analytics for
compliance and Anti Money Laundering (AML)

TERR, Open Source R

Sources:
NY Times : “The Age of Big Data” (Feb 11)
NY Times: “How Companies Learn Your Secrets” (Feb 16)
TIBCO Enterprise Runtime for R (TERR)
• TIBCO owns the S language and have built an R engine
• TIBCO engineers created and support S+
• TIBCO engineers have created the world’s only commercial R
engine: TIBCO Enterprise Runtime for R (TERR)
• TERR built from ground up in C++
• Faster than R; better memory management
• Optimization for known deficiencies in R constructs (e.g. looping)
• Clean room: no code from R
• MKL, lapack, linpack, eispack inside
• Licensing
• IP owned, supported and maintained product
• Scalable, embeddable, redistributable
• TERR is now inside Spotfire and TIBCO Products
• Spotfire Professional
• Spotfire Statistics Services
• TIBCO Business Events, TIBCO Grid Server
Engine Pool
Service dbase
Spotfire
Statistics
Services
TERR
SAS® MATLAB®
C#Java
TERR

Speed Comparison
Model Fitting: 5 Million Rows Model Scoring: 20 Million Rows
TERR

AUDIO & VIDEO IMAGES TEXT WEB & SOCIAL MACHINE LOGS CRM SCM ERP

Data Viz
Challenge

People Who Write Longer Emails Earn Higher Salaries!

Average Email Size
Salaryx10,000

Salaryx10,000
Staff
Senior
Managers

Thank You!
EricNovik
Spotfire IndustryAnalyticsGroup
Twitter:@ericnovik
enovik@tibco.com ::www.ericnovik.com

TIBCO Spotfire: Data Science in the Enterprise

More Related Content

What's hot (20)

Viewers also liked (11)

Similar to TIBCO Spotfire: Data Science in the Enterprise (20)

Recently uploaded (20)

TIBCO Spotfire: Data Science in the Enterprise