FICO Xpress Optimization Suite · FICO ®Precision Marketing Manager FICO® Retail Action Manager...
Transcript of FICO Xpress Optimization Suite · FICO ®Precision Marketing Manager FICO® Retail Action Manager...
Confidential. This presentation is provided for the recipient only and cannot be reproduced or shared without Fair Isaac Corporation's express consent.
© 2011 Fair Isaac Corporation. 1
FICO Xpress Optimization Suite
Oliver BastertSenior ManagerXpress Product Management
September 22 2011
Webinar
© 2011 Fair Isaac Corporation. Confidential.2 © 2011 Fair Isaac Corporation. Confidential.2
Agenda
» Introduction to FICO
» Introduction to FICO Xpress Optimization Suite
» Performance
» Distributed Modelling and Solving
» Case Q&A
© 2011 Fair Isaac Corporation. Confidential.3 © 2011 Fair Isaac Corporation. Confidential.3
Introduction to FICO
© 2011 Fair Isaac Corporation. Confidential.4 © 2011 Fair Isaac Corporation. Confidential.4
Profile
The leader in predictive analytics for decision management
Founded: 1956
NYSE: FICO
Revenues: $605 million (fiscal 2010)
Products and Services
Scores and related analytic models
Analytic applications for risk management, fraud, marketing
Tools for decision management
Clients andMarkets
5,000+ clients in 80 countries
Industry focus: Banking, insurance, retail, health care
Recent Rankings
#1 in services operations analytics (IDC)
#7 in worldwide business analytics software (IDC)
#26 in the FinTech 100 (American Banker)
Offices
20+ offices worldwide, HQ in Minneapolis, USA
2,200 employees
Regional Hubs: San Rafael (CA), New York, London, Birmingham (UK),Munich, Madrid, Sao Paulo, Bangalore, Beijing, Singapore
FICO Snapshot
© 2011 Fair Isaac Corporation. Confidential.5 © 2011 Fair Isaac Corporation. Confidential.5
FICO delivers
superior predictive analytic solutions
that drive
smarter decisions.
Thousands of businesses worldwide,
including 9 of the top Fortune 10,
rely on FICO to
make every decision count.
© 2011 Fair Isaac Corporation. Confidential.6 © 2011 Fair Isaac Corporation. Confidential.6
Transforming Decision Management
Sharpen customer-centric decisions
Predict customer needs and behaviorPinpoint best offer and action
PREDICT
Increase customer profitability
Reduce losses from fraud and riskConnect all decisions about a customer
PROFIT
ADAPT
Change faster and respond flexibly
Change business rules instantlyCreate a test-and-learn culture
IMPROVE
Continually improve strategy performance
Model decisions for greater controlOptimize strategies to grow faster
© 2011 Fair Isaac Corporation. Confidential.7
IMPROVE Strategy Performance
Model decisions for greater control
» Identify the decision drivers and the effects of every action
» Use the decision model as a planning tool to test changes in the business environment
Optimize strategies to grow faster
» Create analytically derived strategies to meet specified business objectives
» Design strategies with millions of variables – instantly
Uses FICO optimization software to
create analytically driven decisions on
fleet distribution and utilization
Deployed across every key market in
continental Europe
• Benefit estimated at $19 million
“FICO™ Xpress tells us, for example:
On Friday morning, bring only four
cars from Heathrow to Mayfair, and
bring another four from Stansted
Airport. The utilization of our fleet has
gone up by one or two percentage
points.”
© 2011 Fair Isaac Corporation. Confidential.8
FICO: Game-Changing Analytics
First commercially available credit scoring systems
First automated origination systems with analytics
First cross-bureau credit scores
First small business scoring systems
First neural network-based fraud solutions
First cardholder profiling for fraud
First insurance underwriting scoring systems
First adaptive control systems for managing card accounts
First credit line optimization solutions
First predictive systems for insurance fraud
First analytic systems for retailers to optimize offers
First adaptive analytics for fraud
First credit capacity scores
First score for prescription adherence
1960s 1970s 1990s1980s 2000s
FICO holds 100+ patents in analytic
and decision management technology,
with 150 more patents pending
© 2011 Fair Isaac Corporation. Confidential.9 © 2011 Fair Isaac Corporation. Confidential.9
Tools Solutions Scores Tools Solutions
Building an Analytic Advantage
Business
Intelligence
Descriptive
Analytics
Predictive
Analytics
Decision
Optimization
Summarize Past and Current Behavior
Automatically
take the ideal
action on each
individual
Target each
decision to a
customer’s future
behavior
Make different
offers to groups
of customers
Understand
the trends in
the business
Predict Future Behavior and Adapt
Decis
ion
Valu
e
Analytic Capability
© 2011 Fair Isaac Corporation. Confidential.10 © 2011 Fair Isaac Corporation. Confidential.10
FICO Product Portfolio
For Lifecycle Specific Decision Processes
Marketing OriginationCustomer
ManagementCollections and
RecoveryFraud
Management
Applications
FICO® Precision Marketing Manager
FICO® Retail Action Manager
FICO®
Origination Manager
FICO® TRIAD®
Customer ManagerFICO® Debt Manager™
FICO™ Recovery Management
System™
FICO™ Falcon®
Fraud Manager
FICO™ Insurance Fraud Manager
For Any Decision Process
Scores
B2B: FICO® Score FICO® Credit Capacity Index™
FICO® Insurance Risk Scores
B2C: myFICO®
Tools
Business Rules Management: FICO™ Blaze Advisor®
Predictive Analytics: FICO™ Model Builder
Optimization: FICO™ Xpress Optimization Suite FICO™ Decision Optimizer
Professional Services
Custom Analytics
Operational Best Practices
Strategy Design and Optimization
© 2011 Fair Isaac Corporation. Confidential.11 © 2011 Fair Isaac Corporation. Confidential.11
Introduction to FICO Xpress Optimization Suite
© 2011 Fair Isaac Corporation. Confidential.12
Xpress Optimization Suite
Solvers
Modelling
Development
De
plo
ym
en
t
LP
MIP
QP
MIQP
QCQP
MIQCQP
SLP
MISLP
NLP
MINLP
CP
MoselMOdelling and Solving Environment Language
XADGraphical user interface development using Mosel
.NET/Java/C/C++/VB
IVEDevelopment Environment
IVE-XADGUI development
Programming
Interfaces
Solver API Mosel API BCL*
GUI
* Builder Component Library for modelling in a programming language
© 2011 Fair Isaac Corporation. Confidential.13
Xpress-IVE: Mosel & Optimizer
» Editor
» Debugger
» Profiler
» Progress graphs
» Visualization
» Wizards
» Mosel extensions
» Deployment
© 2011 Fair Isaac Corporation. Confidential.15
Product Portfolio & Pricing Optimization
FICO Optimization Dashboard: Debt Consolidation ModuleConfidential – do not copy
© 2011 Fair Isaac Corporation. Confidential.18
Key Features and Benefits of Xpress-Mosel
Features Benefits
» Advanced programming languages:
» Algebraic modeling language
» Procedural programming language
» Entire Mathematical Model can be stored in one place for rapid development and easy maintenance.
» Utilize different solvers in the same model
» From Mosel you can solve LPs, MIPs, MIQPs, Non-Linear problems, Stochastic problems, and Constraint problems
» Decompose & parallelize a model to take advantage of multiple CPUs/cores
» Faster solve times
» Make full use of your computing infrastructure through distributed computing
» Build a GUI exclusively within Mosel code
» Decreases development time, gets optimization in front of business user quicker
» Portable across operating systems » Mosel Model compiled in one OS can be deployed on all other supported Operating Systems, decreasing development time
» Open, modular architecture, User extensible
» User flexibility to solve the most complicated optimization problems
» not limited to/by predefined language features
» Compiled » Protects intellectual property
» Offers a variety of APIs and data connectors
» Easy deployment and works in heterogeneous environments
© 2011 Fair Isaac Corporation. Confidential.19
Xpress History and Product Focus
» 26 years of experience in modelling and optimization
» 24 years of experience in mixed integer optimization
» 12 years of experience in nonlinear optimization
» 10 years Xpress-Mosel, modelling and solving environment
» Integration of modelling and solving
» Focus on (potentially) exact solution methods
» Xpress-Solvers often can prove optimality of the solution
» They always give you information on the quality of the solution
© 2011 Fair Isaac Corporation. Confidential.20
Xpress Innovations
» Solving
» 1983: LP solver running on PCs
» 1992: parallel MIP (1997 on distributed PC/Linux networks)
» 1995/1996 : commercial branch and cut algorithm
» 1998: bound switching in dual simplex
» 2003: lift-and-project cuts
» 2009: parallel MIP heuristics
» 2010: LP/MIP solver crosses 64-bit coefficient indexing threshold
» Modelling
» 1983: general purpose algebraic modelling language (mp-model)
» 2001: algebraic modelling language combining modelling, solving, and programming (Mosel)
» 2005: profiler and debugger for a modelling language
» 2005: user-controlled parallelism at the model level
» 2010: algebraic modelling language supporting distributed computing
© 2011 Fair Isaac Corporation. Confidential.21
Xpress differentiators
» Unique capabilities for large scale optimization including ability to solve ultra-large problems (true64bit capabilities) and support for distributed modeling and optimization
» Complete set of state-of-the-art optimization engines that are robust, reliable and faster than competing solutions
» An easy-to-learn, powerful modeling and programming language, Xpress-Mosel
» The premier visual development environment, IVE, for developing mathematical models
» An intuitive drag-and-drop editor for creating GUIs that seamlessly integrate with the model for rapid prototyping and deployment
» A partner committed to solving all of your most difficult optimization problems
© 2011 Fair Isaac Corporation. Confidential.23
Recent enhancements
Xpress 7.1 delivers (GA Nov 2010)
» Solve much bigger problems
» The possibilities are limitless with the enhanced optimization and modeling support for solving ultra-large-scale models where the number of coefficients can exceed 2 billion.
» Solve large problems faster
» Cut solution times dramatically by leveraging distributed execution of Mosel models that can now be controlled from a master model across a heterogeneous set of machines.
» Get solutions faster with significantly improved solver performance
» Average increase of 50% arithmetic/50% geometric for multi-threaded MIP
» Average increase of 50% arithmetic/25% geometric for single-threaded MIP
» Average increase of 50% arithmetic/70% geometric for MIQP
» Significant improvements to speed and stability of quadratic simplex and SLP non-linear algorithms
© 2011 Fair Isaac Corporation. Confidential.24
Recent enhancements
» Improved developer usability
» Developers will also enjoy greater productivity from usability improvements to the development environment and enhanced modeling functionality such as the MIIS automated modeling error/infeasibility detection.
» Easier to integrate with other applications
» Optimization programmers will be pleased by the addition of simplified and more robust data exchange capabilities between Mosel and applications.
Xpress 7.2 delivers (GA April 2011)
» Exceptional public benchmark performance
© 2011 Fair Isaac Corporation. Confidential.25 © 2011 Fair Isaac Corporation. Confidential.25
Performance
© 2011 Fair Isaac Corporation. Confidential.26
Comparing Solver Performance
» Solver performance is important but not the only decision criterion
» Selection of benchmark sets
» Represent client mix of problems
» Solvable instances but not too simple
» Feasibility and optimality check of solution
» Numerically stable problems are preferred for performance benchmarks
» The only public benchmarks for optimization solvers is run by Hans Mittelmann. He frequently changes the benchmarking sets
» The best known collection of MIP instances is currently updated from version MIPLIB 2003 to version MIPLIB 2010 and will contain for the first time an agreed benchmarking subset.
» A benchmark comparison is always a snapshot of the performance of the available software at a given point in time.
© 2011 Fair Isaac Corporation. Confidential.27
The MIPLIB 2003 Experience
ProblemOld Best Known Obj.
Value (*)Xpress Improved Obj.
Value (**)GAIN
(|1-(**)/(*)|)
atlanta-ip 95.009549704 90.00987861 5.3%
msc98-ip 20980991.006 19839497.006 5.4%
protfold -30 -31 3.3%
rd-rplusc-21 171182 165395.2753 3.4%
sp97ar 664565103.76 660705646.5 0.6%
stp3d unknown 500.736 N/A
ds 283.4425 116.59 58.9%
momentum3 370177.036 236426.335 36.1%
t1717 193221 170195 11.9%
liu 1172 1102 5.9%
dano3mip 691.2 687.733333 0.5%
Op
tim
al
Un
so
lved
Solving Hard Mixed Integer Programming Problems with Xpress-MP:
A MIPLIB 2003 Case Study, Informs Journal on Computing, 2009
by Richard Laundy, Michael Perregaard, Gabriel Tavares, Horia Tipi, and Alkis Vazacopoulos
© 2011 Fair Isaac Corporation. Confidential.28
Geometric Mean is the comparison criterion of choice
» Instead of comparing the overall runtime on a given matrix set (or equivalently the average runtime or arithmetic mean on that set) the accepted way of comparing optimization solver performance is by comparing the Geometric Mean
» The presence of a few extremely small or large values has no considerable effect on geometric mean so it measures performance more accurately than arithmetic mean which is biased towards large outliers.
» The geometric mean denotes the most likely runtime you will observe for an instance of the test set.
© 2011 Fair Isaac Corporation. Confidential.29
Standard LP Problems (Barrier, Simplex)Public Benchmark by H. Mittelmann as of 29 Apr 2011, GeoMean
0
10
20
30
40
50
60
70
80
90
XPRESS CPLEX GUROBI MOSEK
Standard LP Problems (Barrier, Simplex)
© 2011 Fair Isaac Corporation. Confidential.30
Barrier on Large LP ProblemsPublic Benchmark by H. Mittelmann as of 29 Apr 2011, GeoMean
0
100
200
300
400
500
600
700
XPRESS CPLEX GUROBI MOSEK
Barrier on Large LP Problems
© 2011 Fair Isaac Corporation. Confidential.31
MIQP ProblemsPublic Benchmark by H. Mittelmann as of 29 Apr 2011, GeoMean
0
50
100
150
200
250
300
350
400
450
XPRESS CPLEX GUROBI SCIP
MIQP Problems
© 2011 Fair Isaac Corporation. Confidential.32
MIPLIB 2010 Benchmark setGeometric Mean, single threaded, MIPLIB 2010 paper
0.00
50.00
100.00
150.00
200.00
250.00
300.00
XPRESS CPLEX GUROBI
© 2011 Fair Isaac Corporation. Confidential.33 © 2011 Fair Isaac Corporation. Confidential.33
Distributed Modelling and Solving
c©2011 Fair Issac Corporation.
Schemes of parallelization
1. Simple submodel run
wait fortermination
compile/load/runsubmodel
process resultsresults
start
results
Submodel
Master
User
c©2011 Fair Issac Corporation.
Schemes of parallelization
2. Iterative sequential submodel runs(decomposition algorithms)
process results
wait fortermination
start
run submodel
compile/loadsubmodel
results
results
Submodel
Master
User
c©2011 Fair Issac Corporation.
Schemes of parallelization
3. Independent parallel submodels
wait fortermination
compile/load/runsubmodels
process resultsresults
start
results
start
...
Master
User Submodel 1 Submodel n
c©2011 Fair Issac Corporation.
Schemes of parallelization
4. Communicating concurrent submodels
wait for events
process events
process results
startstart
...
compile/load/runsubmodels
results
broadcast updates/termination
events/results
Master
Submodel 1 Submodel n
User
c©2011 Fair Issac Corporation.
Distributed solving
» Use all the computing power available in yourlocal network by solving (sub)models onremote machines
run master model
Local
c©2011 Fair Issac Corporation.
Distributed solving
» Use all the computing power available in yourlocal network by solving (sub)models onremote machines
run master model run submodel
Local Remote
c©2011 Fair Issac Corporation.
Distributed solving
» Use all the computing power available in yourlocal network by solving (sub)models onremote machines
run master model run submodel
Local Remote
» Physical location of model files, input andresult data depending on application
c©2011 Fair Issac Corporation.
Distributed solving:Location of models and data
1. On local host» physical files or in memory (e.g. included in master
model or in calling host application)» a (master) model can recursively start new instances
of itself
submodel file
data, results
run master model run submodel
load
Local Remote
c©2011 Fair Issac Corporation.
Distributed solving:Location of models and data
2. On remote host» configurable read/write access on remote machine
submodel file
data, results
run master model run submodel
load
Local Remote
c©2011 Fair Issac Corporation.
Distributed solving:Location of models and data
3. Centralised repository» eases version control in multi-user environments
submodel file
data, results
run master model run submodel
load
Local Remote
Central repository
c©2011 Fair Issac Corporation.
Distributed applications
» Multi-user application with Mosel model asdispatcher
Mosel server
User
UserD
atab
ase
Productionmachine
machineProduction
... ...
» Example: optimization applications in finance(solving large numbers of small to mediumsize problems)
c©2011 Fair Issac Corporation.
Distributed applications
» Decomposition with central data store
Database
User ...Moseloptimization master
Submodel
Submodel
» Examples: Column generation in transport orpersonnel planning; blockwise(Dantzig-Wolfe) decomposition in productionplanning
c©2011 Fair Issac Corporation.
Distributed applications
» Decomposition with remote, distributed datasources
Dat
aD
ata
User ...Moseloptimization master
Remotemodel
Remotemodel
» Example: Large-scale planning inheterogeneous computing environment
© 2011 Fair Isaac Corporation. Confidential.8 © 2011 Fair Isaac Corporation. Confidential.8
Case Studies
c©2010 Fair Issac Corporation.
Portfolio rebalancing:Problem description
» Modify the composition of an investmentportfolio as to achieve or approach a specifiedinvestment profile.
c©2010 Fair Issac Corporation.
Optimization application in MoselStandalone
Data files Mosel model
IVE
Output files
start applicationreturn results
c©2010 Fair Issac Corporation.
Optimization application in MoselXAD GUI
Data files Mosel model
XAD
Output files
outputSummary
ration fileConfigu-
start applicationreturn results
c©2010 Fair Issac Corporation.
Optimization application in MoselEmbedded into host application
Mosel model
Output files
outputSummary start application
return results
JavaData files
c©2010 Fair Issac Corporation.
Optimization application in MoselAlternative interfaces
outputSummary
Data files
start application
Mosel model
XAD IVE Java
Output files
outputSummary
Data filesration fileConfigu-
return resultsoutputSummary
c©2010 Fair Issac Corporation.
Some highlights
» Model:» easy maintenance through single model» deployment as BIM file: no changes to model by
end-user» language extensions according to specific needs
» Interfaces:» several run modes adapted to different types of
usages» efficient data exchange with host application
through memory» parallel model runs (Java) or repeated sequential runs
(XAD)
c©2010 Fair Issac Corporation.
Aircraft routing:Problem description
» For given sets of flights and aircraft,determine which aircraft services a flight.
» Aircraft are not identical» they cannot all service every flight» a specific maintenance site must be used per plane» some scheduled long maintenance breaks
» Starting condition: each aircraft has a startingposition and a specific amount of accumulatedflight minutes
c©2010 Fair Issac Corporation.
Aircraft routing:Application architecture
» Master problem: route selection» Subproblems: route generation (one instance
per plane)» parallel, possibly remote, execution of submodels
» User interface (optional): XAD GUI