Analytics Julyaugust 2012

Published on January 2017 | Categories: Documents | Downloads: 52 | Comments: 0 | Views: 425
of 47
Download PDF   Embed   Report

Comments

Content


h t t p : / / www. a n a l y t i c s - maga z i n e . O R g
ALSO INSIDE:
July/ august 2012 DRiving BetteR Business DecisiOns
• The Future of Forecasting
intersection of analytics and insight,
plus survey of software packages
• Analytics & Big Data
skeptics vs. enthusiasts: Battle for
hearts and minds of decision-makers
• Analytics at IBM
technology, consulting giant’s long,
rich history with advanced analytics
COMpETITIvE
CrOwDSOurCINg
ANALyTICS AS SpOrT:
A wIN-wIN FOr
CONTESTANTS AND
COrpOrATIONS
executive edge
aryng ceO
piyanka Jain
on what predictive
analytics ‘is not’
Bellevue, WA, USA • Haarlem, the Netherlands • Singapore • Shanghai, China
www.aimms.com • [email protected]
AIMMS is a registered trademark of Paragon Decision Technology B.V. Other brands and their products are trademarks of their respective holders
The AIMMS PRO Platform
A New Revolution In Optimization
Ever thought about
... creating an optimization App Store
for business users?
... providing web access to your
optimization solutions?
...deploying in the cloud?
No? We have!
Check out our AIMMS PRO Platform at:
www.aimms.com/aimms-pro-platform
04888-AIMMS-ORMS.indd 1 20-06-12 13:37
1 | a na ly t i cs - maga z i ne . or g
the sport of data science
i ns i De s t ORy
How many analysts does it take to
solve a problem?
That may sound like the start of a bad
joke, but no one was laughing in 2006
when Netfix offered $1 million to any-
one who could come up with a collabora-
tive fltering algorithm that improved the
performance of Cinematch (Netfix’s in-
house software) by at least 10 percent.
Cinematch predicts which movies Netfix
customers like and makes movie rec-
ommendations to customers based on
those predictions. The goal: boost cus-
tomer satisfaction and retention along
with sales.
Three years later, after receiving sev-
eral thousand entries from more than 100
countries, a winner was announced, the
$1 million prize was awarded and a cot-
tage industry – online marketplaces for
business projects where companies post
challenges, provide data and offer prizes
for the best solutions – took off.
While Netfix reportedly performed
no formal cost-beneft analysis on the
Netfix Prize, the company was clearly
pleased with the results. At the time,
Netfix CEO Reed Hastings said, “You
look at the cumulative hours and you’re
getting Ph.D.’s for a dollar an hour.”
In this month’s cover story, Margit
Zwemer, a data scientist and commu-
nity manager at Kaggle, provides an
inside look at “crowdsourcing” – the
concept that turns complex analytical
problems into a competitive sport open
to analysts, astrophysicists or anyone
else who cares to submit a solution. As
Zwemer notes in her article, the con-
cept is not new; as far back as the 18th
century, the British government offered
more than £100,000 in prize money to
anyone who could come up with simple
and practical methods for measuring
longitude to assist maritime navigation.
The Netfix Prize, however, helped
turn crowdsourcing into a modern-day,
mainstream corporate strategy. “Data
research competitions are a resource-
effcient way for organizations to solve
complex data problems, and they cre-
ate a meritocratic market for talent that
changes the way analysts work,” Zwe-
mer writes. Kaggle, an online platform
for predictive modeling and analytics
competitions, was one of many compa-
nies that jumped on the “crowdsourcing”
bandwagon in the aftermath of the Net-
fix Prize. According to Zwemer, Kaggle
boasts a worldwide online community
of more than 40,000 data scientists and
predictive analysts, competing under the
slogan “making data science a sport.” ❙
– peteR hORneR, eDitOR
[email protected]
DRIVING BETTER BUSINESS DECISIONS
C o n t e n t s
FEATurES
ReseaRch as a cOmpetitive spORt
By Margit Zwemer
Crowdsourcing the full analytics value chain, starting with the
predictive model itself.
cOllaBORative fORecasting: fROm visiOn tO
Reality
By Brian Lewis
transitioning from periodic, isolated activities to a single, real-time
enterprise process.
fORecasting an upwaRD tRenD?
By Jack Yurkiewicz
survey of forecasting software reveals interesting trends and
new developments.
analytics & Big Data: skeptics vs. enthusiasts
By Gary Cokins
Acceptance of analytics to tackle issues related to big data
going through growing pains.
pReDictaBility Of time seRies
By Subir Mansukhani
statistical measure to infer the level of diffculty in choosing an
appropriate model.
cORpORate pROfile: iBm
By Arnold Greenland
technology, consulting giant boasts a long, rich history with
advanced analytics.
12
16
20
25
29
32
2 | a na ly t i cs - maga z i ne . or g
Discover the easiest, fastest, lowest cost and risk “on
ramp” to full-power data mining and predictve
analytcs: Analyze your data in Microsof Excel with
XLMiner sofware, now from Frontline Systems.
You won’t need a statstcs PhD, months of learning
tme, or a huge budget for “enterprise data mining”
sofware. But you will have the power of predictve
analytcs and data visualizaton at your fngertps.
Sophistcated Data Mining Power.
XLMiner goes far beyond other statstcs and forecastng
add-ins for Excel. It starts with multple regression,
exponental smoothing, and ARIMA models, but goes
further with regression trees, k-nearest neighbors, and
neural networks for predicton. It ofers discriminant
analysis, logistc regression, k-nearest neighbors,
classifcaton trees, naïve Bayes and neural nets for
classifcaton, and associaton rules for afnity analysis.
Practcal Data Cleaning and Exploraton.
XLMiner includes utlites for data sampling and
parttoning, missing data handling, binning, and
transforming categorical data. You can apply principal
component analysis, k-means clustering, and
hierarchical clustering to simplify and cluster your data.
Help, Textbook and Instructor Support.
XLMiner’s extensive online Help introduces the data
mining methods and explains the optons in each dialog.
With the popular textbook Data Mining for Business
Intelligence – built around XLMiner sofware – you can
go from beginner to expert business analyst in a high-
demand feld. Instructors can take advantage of low-
cost classroom licenses, PowerPoint slides, example
fles, solutons to exercises, an online forum, and more.
Download Your Free Trial Now.
Visit www.solver.com/xlminer to learn more, register
and download a free trial – or email or call us today.
Tel 775 831 0300 • Fax 775 831 0314 • [email protected]
XLMINER: VISUAL DATA MINING
Works with PowerPivot to Access 100 Million Rows in Excel
25
20
12
J ULY/ AUGUst 2012
Brought to you by
DRIVING BETTER BUSINESS DECISIONS
Brought to you by
The Excel Solver’s Big Brother Makes Optmizaton
Easier, More Powerful – and Far More Visual.
From its classic Solver Parameters dialog to its Ribbon,
Task Pane, charts to visualize the shape of functons,
Constraint Wizard, model diagnosis, Help, and coverage
in leading management science textbooks, Risk Solver
Platorm makes it easier to learn and use optmizaton.
Learn and Use Monte Carlo Simulaton, Decision
Analysis, and More with the Same Sofware.
You can use its powerful optmizaton features alone –
but Risk Solver Platorm is also a full-power tool for
Monte Carlo simulaton and decision trees, with a Dis-
tributon Wizard, 50 distributons, 30 statstcs and risk
measures, and a wide array of charts and graphs.
Use the Full Spectrum of Optmizaton Methods.
Fast linear, quadratc and mixed-integer programming is
just the startng point in Risk Solver Platorm. SOCP,
nonlinear, non-smooth and global optmizaton are just
the next step. Easily incorporate uncertainty and solve
with simulaton optmizaton, stochastc programming,
and robust optmizaton – all at your fngertps.
Go Beyond Single Optmizaton/Simulaton Solutons.
Defne parameters in seconds, run high speed mult-way
parameterized simulatons and optmizatons, capture
all the solutons, and produce instant reports, charts
and graphs, without programming. And you can do
even more in Excel VBA.
Free Trials and Support for Your Eforts at Solver.com.
Download a free trial of Risk Solver Platorm, and learn
about the many other ways we can support your
learning and teaching eforts – all at www.solver.com.
TEACH AND LEARN OPTIMIZATION AND SIMULATION
RISK SOLVER PLATFORM
Tel 775 831 0300 • Fax 775 831 0314 • [email protected]
3 | a na ly t i cs - maga z i ne . or g
REGISTER FOR A FREE SUBSCRIPTION:
http://analytics.informs.org
INFORMS BOARD OF DIRECTORS
President Terry P. Harrison, Penn State University
President-Elect Anne G. Robinson, Verizon Wireless
Past President Rina R. Schneur,
Verizon Network & Technology
Secretary Brian Denton,
North Carolina State University
Treasurer Nicholas G. Hall, Ohio State University
Vice President-Meetings William “Bill” Klimack, Chevron
Vice President-Publications Linda Argote, Carnegie Mellon University
Vice President-
Sections and Societies Barrett Thomas, University of Iowa
Vice President-
Information Technology Bjarni Kristjansson, Maximal Software
Vice President-Practice Activities Jack Levis, UPS
Vice President-International Activities Jionghua “Judy” Jin, Univ. of Michigan
Vice President-Membership
and Professional Recognition Ozlem Ergun, Georgia Tech
Vice President-Education Joel Sokol, Georgia Tech
Vice President-Marketing,
Communications and Outreach E. Andrew “Andy” Boyd,
University of Houston
Vice President-Chapters/Fora Olga Raskina, Con-way Freight
INFORMS OFFICES
www.informs.org • Tel: 1-800-4INFORMS

Executive Director Melissa Moore
Meetings Director Teresa V. Cryan
Marketing Director Gary Bennett
Communications Director Barry List

Headquarters INFORMS (Maryland)
7240 Parkway Drive, Suite 300
Hanover, MD 21076 USA
Tel.: 443.757.3500
E-mail: [email protected]
ANALYTICS EDITORIAL AND ADVERTISING
Lionheart Publishing Inc., 506 Roswell Street, Suite 220, Marietta, GA 30060 USA
Tel.: 770.431.0867 • Fax: 770.432.6969
President & Advertising Sales John Llewellyn
[email protected]
Tel.: 770.431.0867, ext.209
Editor Peter R. Horner
[email protected]
Tel.: 770.587.3172
Art Director Lindsay Sport
[email protected]
Tel.: 770.431.0867, ext.223
Advertising Sales Sharon Baker
[email protected]
Tel.: 813.852.9942
the Institute for Operations Research and the Management
Sciences (INFORMS). For a free subscription, register at
http://analytics.informs.org. Address other correspondence to
the editor, Peter Horner, [email protected]. The
opinions expressed in Analytics are those of the authors, and
do not necessarily refect the opinions of INFORMS, its offcers,
Lionheart Publishing Inc. or the editorial staff of Analytics.
Analytics copyright ©2012 by the Institute for Operations
Research and the Management Sciences. All rights reserved.
DEpArTMENTS
inside story
executive edge
proft center
analyze this!
forum
five-minute analyst
thinking analytically
conference preview
learning Resources
viewpoint
1
4
6
7
10
35
37
41
43
45
37
41
www. i nf or ms . or g
Predictive analytics is a powerful method
used by leading business organizations to pre-
dict future events and behavior in order to op-
timize current marketing, product, operations
and sales actions. The prediction is based on
a basic fundamental: “past behavior predicts
future behavior.”
As an example, let’s take an e-commerce
company selling furniture online. The online
site has about a 10 percent contact rate, i.e.,
of the 100 people who come to the website,
about 10 people contact customer service
within 24 hours. The VP of Customer Service
Operations is looking to reduce the contact
rate as each contact costs the company incre-
mental dollars in operations. Let’s say a 1 per-
cent decrease in contact rate would amount to
a savings of $1 million a year for the company.
The operations team decides to use predictive
analytics to understand the drivers of contact.
By using certain predictive analytics tech-
niques on historical data, a relationship is
identifed between help page visits and visitors
making a phone call to the customer service
department. Specifcally, 50 percent of visi-
tors, after having gone through three distinct
help pages, call customer service. This is a
very helpful clue. If the visitor can be inter-
cepted before he hits the third help page either
by providing better help content or a live chat
or a clarifcation window based on what he is
browsing, contact rate can likely be reduced. I
have seen organizations save and make mil-
lions by understanding this kind of pivotal rela-
tionships between behaviors and events.
On the fip side, predictive analytics, in spite
of being a powerful optimization technique, is
often left to the devices of data miners and
data scientists, and thus is often misunder-
stood and misused by businesses. Having a
frm hold on what predictive analytics is not will
make predictive analytics a more useful tool
for businesses.
1. Predictive analytics is not new. Some
news items such as the Feb. 16 New York
Times report on retail company Target’s predic-
tion of teenage pregnancy implies that predic-
tive analytics is a new found technique. But in
fact predictive analytics is not new. Fischer and
Durand, founders of the Econometric society,
built one of the frst credit scoring models 80
years ago. But predictive modeling techniques
go back thousands of years –Indian astrologi-
cal charts used to arrange marriages are one
such example.
2. Predictive analytics does not produce
perfect predictions. Often while building the
model, it is clear to all that model prediction has
a probability associated with it, but upon suc-
cessful use, there is often a misplaced sense
of perfectness in the scores. As in the case of
our e-commerce company, 50 percent of visi-
tors after visiting three help pages are going
to call, but the other 50 percent won’t call. The
model predicts by maximizing the likelihood,
and a certain degree of misclassifcation al-
ways exists. By using other predictors, these
odds can be improved, but the prediction will
still not be a 100-percent accurate.
3. A good software tool does not mean a
good model. With tremendous develop-
ment on the software tool front with better GUI
as well as higher automation, people new to
the feld often mistakenly believe that a good
model can be automatically built by pressing
a “build model” button. A good model requires
proper technical skills and a proper model-
building process. Surprisingly, sometimes
even proper skills and a proper process does
not deliver a good model with decent lift and
low misclassifcation.
4. A good model does not always mean
better business results. This is one of those
highly prevalent myths that even experienced
analysts fall for, leaving them frustrated when
nobody in the business seems to care for the
amazing model they built. A good model gen-
erates business impact only when the right
stakeholders are brought into the analytics
process at the right time, thus building proper
4 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
what predictive analytics
“is not”
By piyanka Jain
predictive analytics, in
spite of being a powerful
optimization technique, is
often left to the devices
of data miners and data
scientists, and thus is
often misunderstood and
misused by businesses.
e Xe cu t i ve e D ge
5 | a na ly t i cs - maga z i ne . or g
alignment toward actionability using a
framework. In our e-commerce exam-
ple, if instead of building a contact rate
prediction model, we had built a model
to predict visitors most likely to do live-
chat, would the VP of Operations care
for the model and the results? Prob-
ably not. Unless we can show the re-
lationship between what we (predictive
model/team) are trying to do and his
business goal (contact rate reduction),
the VP of Operations is not going to use
the model.
5. Models can’t be built and forgot-
ten. Models become stale over time.
If not maintained, they often stop de-
livering the incremental value it start-
ed with. As organizations embark on
the journey of competing on analytics,
they need to be aware that it is not a
one-time investment. You can’t hire ex-
ternal consultants, get the model built
and leave it at that. Model needs to be
tested, tweaked and then maintained
to continue delivering the incremental
beneft. ❙
Piyanka Jain ([email protected]) founder, president
and CEO of analytics training company Aryng, speaks
regularly at business and analytics conferences on
data-driven decision-making in an organization. Her
prior roles include head of NA Business Analytics at
PayPal and senior marketing analytics position with
Adobe. She is an INFORMS partner.
e Xe cu t i ve e D ge
The Optmizaton / Simulaton Library that Speaks the
Languages of Desktop and Web Computng.
No other “callable library” matches the fexibility of
Solver SDK Platorm – from solving Excel models on
your server, to solving models created in your browser.
Build Your Model Faster in Excel or Visual Studio.
Use Risk Solver Platorm or its subsets to build your
model in Excel. Use Microsof Visual Studio to create
your model in C++, C# or VB.NET. Use C, R, MATLAB,
Java, PHP or JavaScript – it’s your choice.
Use Optmizaton and Simulaton as a Web Service.
Run the Solver Server that comes with Solver SDK on
any PC or server, then solve your models “over the
wire” using our standard Web Service, SOAP and WSDL
protocols. Get real-tme progress, or just disconnect
and retrieve the soluton later – it’s your choice.
Solve Your Most Challenging Linear Mixed-Integer,
Nonlinear and Stochastc Optmizaton Models.
Solver SDK’s wide choice of built-in and plug-in Solver
Engines handle every type and size of optmizaton
problem, from Gurobi and Xpress-MP to Mosek and
KNITRO. Use the SDK’s built-in simulaton optmizaton
and stochastc programming algorithms, and support
for mult-core CPUs.
Download a Free Trial at www.solver.com.
Learn more and download a FREE trial version of Risk
Solver Platorm, Solver SDK Platorm, and our eight
plug-in Solver Engines from Solver.com
THE MOST FLEXIBLE WAY TO DEPLOY YOUR MODEL
SOLVER SDK PLATFORM
Tel 775 831 0300 • Fax 775 831 0314 • [email protected]
models become stale over
time. if not maintained,
they often stop delivering
the incremental value
it started with. as
organizations embark on
the journey of competing
on analytics, they need to
be aware that it is not a
one-time investment.
subscri be t o Anal yt i cs
It’s fast, it’s easy and it’s FREE!
Just visit: http://analytics.informs.org/
www. i nf or ms . or g
In the 2011 March/April issue of Analytics I
took a look at the question, “What is analytics?”
I shied away from proposing a defnition for a
host of reasons, many of which are illustrated
in the defnition of another discipline with a
long and distinguished history: mathematics.
So what is “mathematics”?
Thanks to our educational system, we all
have exposure to math from an early age. Most
of us associate it with numbers. Ask mathema-
ticians, however, and they may not even men-
tion numbers, focusing instead on abstract
symbol manipulation and sets. Yet even here
there’s no uniform agreement. Consider this
rather striking passage from the Wikipedia
page Defnitions of Mathematics: “Mathematics
has no generally accepted defnition. Different
schools of thought, particularly in philosophy,
have put forth radically different defnitions. All
are controversial.”
Nonetheless, we speak cogently about
mathematics, and most people would argue
that mathematics is important, leading us to
conclude that an exact defnition isn’t a pre-
requisite for a feld to stand on its own.
Still, the process of seeking a defnition is
invaluable. In my earlier article I focused on
the fact that discussing what analytics is and
isn’t helps us come to a better understanding
of it even if we never converge on a precise
defnition that all businesses, vendors and ac-
ademics can agree on. Here I’d like to focus
on another benefcial aspect of seeking a def-
nition: communicating one party’s perspective
on analytics to the rest of the world. When
John says analytics, what does he mean?
INFORMS (the professional society that pub-
lishes this magazine; see “Proft Center,” March/
April 2012) recently undertook an extensive ef-
fort to arrive at its own defnition of analytics.
The process was spearheaded by the board of
directors, and the goal was to succinctly com-
municate INFORMS’ perspective on the topic.
Needless to say, getting an organization of more
than 10,000 members to agree on anything is a
challenge. And, not surprisingly, not every indi-
vidual member was personally contacted. An ad
hoc committee reviewed existing defnitions, dis-
cussed these defnitions and prepared its own. It
then sent a proposal to various subgroups with-
in INFORMS which in turn solicited comments
from their membership. Many strong, thoughtful
opinions were voiced. But in a matter of months,
a reasonable consensus had emerged. This was
the result:
Analytics is the scientifc process of trans-
forming data into insight for making better
decisions.
At the heart of the defnition is data, and
transforming that data into insight. Of the many
defnitions of analytics now in use, data is cen-
tral to most, as is using that data for a specifc
purpose. Interestingly, the INFORMS defnition
describes that purpose as insight for making
better decisions. INFORMS tends toward the
mathematical end of the analytics spectrum,
so a more technical term than insight wouldn’t
have been surprising. The choice of insight
communicates that while the organization be-
lieves math is important, it’s subsidiary to the
insight aimed at making better decisions.
Even more interestingly in the defnition is
the focus on the scientifc process of transform-
ing data. Is this where mathematics enters the
equation? Not necessarily. Mathematical tools
can be very useful, but the explicit statement in
the defnition calls out the scientifc process, a
process that involves an informed, logical, or-
derly sequence of steps. The scientifc process
is powerful and is something that’s all too fre-
quently overlooked in decision-making. Coupled
with data, the scientifc process is a juggernaut.
Aspects of INFORMS’ defnition aren’t
unique. All defnitions share numerous com-
monalities, which is reassuring. Where the def-
initions tend to diverge is over how much and
what type of mathematics is used. Is simple
reporting part of analytics? It depends upon
whom you ask.
INFORMS’ defnition certainly doesn’t re-
solve the question, “What is analytics?” It
does, however, provide a thoughtful perspec-
tive to ponder. And it informs us of just where
INFORMS stands. ❙
Andrew Boyd, senior INFORMS member and INFORMS VP of
Marketing, Communications and Outreach, has been an executive
and chief scientist at an analytics frm for many years. He can be
reached at [email protected].
6 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
Revisiting ‘what is
analytics’
By e. anDRew BOyD
analytics is the scientific
process of transforming
data into insight for
making better decisions.
p R Of i t ce nt e R
www. i nf or ms . or g
Been cleaning out the home offce this
month. Apparently, this is the kind of thing that
is done by a newly tenured professor with no
summer teaching or administrative responsi-
bilities to avoid tackling the gnarly research
problems on his desk. In addition to the many
long-lost letters, books, papers, ticket stubs,
non-disclosure agreements, unmarked CDs,
forgotten photos, abandoned textbooks and
stray postcards, I’ve also managed to un-
earth an impressive pile of discarded plas-
tic lanyards, the kind one acquires at events
such as orientations, new student dinners and
conferences.
In particular, the bright blue lanyards from
the INFORMS Conference on Business Analyt-
ics and Operations Research (BAOR) in 2011
and 2012 caught my eye. I’ve been going to
the fall INFORMS conference for many, many
years [1], and it’s always a great pleasure to
catch up with familiar old friends there. But at
these past two BAOR conferences, the frst
ones that I have attended, I have also really
enjoyed meeting interesting people who are
new to INFORMS. At the 2011 BAOR confer-
ence in Chicago, I had my frst contact with an
analytics leader named Scott Friesen. At the
2012 BAOR conference in Huntington Beach,
Calif., I had a really interesting breakfast con-
versation with an auto industry pioneer named
Sandy Stojkovski.
Inspired by the discovery of these old lan-
yards, I recently caught up with Scott and
Sandy last week to answer a couple of simple
questions: “Who are you?” and “How did you
end up attending an INFORMS conference?”
Their responses were both fascinating and
instructive.
Scott Friesen is senior director of analytics
within the Customer Insights Unit (CIU) at Best
Buy. This is the latest in a series of roles that
he has played within that company since fnish-
ing his MBA in 2004. Over the past two years,
Scott’s group has grown from 11 to more than
30 professionals by consistently demonstrat-
ing business value to its internal customers.
Scott’s CIU is viewed within Best Buy as the go-
to place for executives and managers across
the company to understand customer behav-
ior, proftability models, competitive trends and
a number of other data-driven questions of in-
terest to executives, marketers, retail manag-
ers and strategic planners.
The journey that led him to this role as an
analytics leader has been unique and can be
traced at least as far back as his business
school statistics class with Paul Glasserman
at Columbia (he confessed to me that he
found Professor Glasserman and his course
to be “inspiring”). Along the way, he’s worked
in strategy for BestBuy.com, been trained as
Six Sigma Blackbelt and Master Blackbelt,
launched a new services business within the
Geek Squad organization, and set up a claims
analytics function (complete with actuaries) to
manage risk in the company’s extended ser-
vice warranty business.
Sandy Stojkovski is president of Scenaria
(www.scenaria.com), a consulting frm that she
founded with AVL (www.avl.com) in 2010. Sce-
naria develops sophisticated analytic models
to provide unique insights and strategy guid-
ance to its clients, who are typically executives
within the auto industry. Starting this company
is only the most recent time that Stojkovski
has “broken the mold” during her career in the
auto industry. She has also been a 22-year-
old greenfeld plant manager, gained exper-
tise in value engineering while working for a
parts manufacturer, and developed a global
power train strategy for a prominent automo-
tive brand. She also managed to complete a
graduate engineering degree and an MBA be-
fore turning 30 while working full-time.
Prior to starting Scenaria, Stojkovski de-
veloped deep expertise with vehicle fuel ef-
ficiency standards first while working on fuel
economy as an engineer at Ford and then
while launching and growing an energy ef-
ficiency practice at an engineering services
firm called Ricardo. Her bold vision – “to trans-
form the way that auto company executives
do strategic planning” – has in turn required
the company to develop a management
7 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
Business analytics
conference builds
connections
By viJay mehROtRa
networking at infORms
conference on Business
analytics and Operations
Research pays dividends.
analy z e t h i s !
www. i nf or ms . or g 8 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
science-based approach to manage-
ment consulting to help its clients intelli-
gently address the types of high-stakes
decisions that have to be made in the
capital-intensive automotive industry.
The company’s projects typically fea-
ture a variety of analytic tools, includ-
ing experimental design and simulation
modeling and visualization.
At frst, I found it curious that neither
Scott nor Sandy had any degrees in math-
ematics, statistics or operations research
(though each of them appreciated what they
had picked up in their respective academic
programs). While both expressed genuine
respect for the mathematical techniques
underpinning this analytics revolution, nei-
ther felt that their relative lack of formal
mathematical training had handicapped
them professionally in any way.
Instead, what really stood out was that
each of them had an almost viscerally
strong inclination toward systems think-
ing. While Friesen attributed this to his
undergraduate roots in the liberal arts,
Stojkovski credits her engineering back-
ground for this orientation toward seeing
problems in the context of the larger sys-
tems in which they exist.
They also shared several other key
traits that are typical of successful leaders
in the analytics industry. Throughout their
careers, the interesting variety of roles and
assignments revealed both of them to be
voracious learners, as well as smart, judi-
cious risk takers. Friesen and Stojkovski
also both emphasized the value of building
strong relationships, with Friesen passing
along an important equation that a mentor
had shared with him (“Trust = Credibility *
Intimacy/Risk”). Each of them repeatedly
spoke of looking at problems strategically
and continually seeking to get closer to the
root of the problem because, as Friesen put
it, “that’s where the leverage is.”
This search for valued relationships
and intellectual leverage is what led both
Friesen and Stojkovski to INFORMS and
in particular to the increasingly popu-
lar spring BAOR conferences. At the
2011 BAOR conference, Friesen made
a presentation [2] about leveraging tech-
nology and relationships to implement
analytic capabilities in the complex and
dynamic world of Best Buy, participated
in a lively panel session on “Growing an
Analytics Capability,” and enjoyed the
analy z e t h i s !
Join the Community
for inside information.
SECTION ON
ANALYTICS
Looking to interact and network
with analytics professionals?
http://analytics.section.informs.org
while both expressed
genuine respect for the
mathematical techniques
underpinning this analytics
revolution, neither felt that
their relative lack of formal
mathematical training
had handicapped them
professionally in any way.
hel p promote Anal yt i cs
It’s fast and it’s easy! Visit:
http://analytics.informs.org/button.html
9 | a na ly t i cs - maga z i ne . or g
mix of industry and academic conver-
sations. Meanwhile, Stojkovski started
out reading Analytics magazine as part
of her ongoing search for new knowl-
edge and became intrigued with the
opportunity to leverage lessons from
different industries into Scenaria’s ap-
proaches. Eventually she made her
way to the 2012 BAOR conference in
Huntington Beach, where she delivered
a presentation [3] on a strategic plan-
ning, analysis and a data visualization
platform that her company had devel-
oped for the California Hybrid, Effcient
and Advanced Truck Research Center
to help the organization to jointly evalu-
ate technology and policy decisions.
These days, there is a great deal of
general talk about the rapid growth and
evolution of analytics. Friesen and Stojk-
ovski provide two specifc and optimistic
visions for the future. Scott and his group
have recently received management ap-
proval for a new multi-million dollar cus-
tomer information infrastructure that he
believes will lead to a quantum growth in
the types of insights his group can pro-
vide to its many business customers with-
in Best Buy. For Sandy, the progressive
tightening in the CAFE standards, which
are slated to increase by 80 percent by
2025, means complex investment deci-
sions and serious risks for the automotive
industry – and great opportunities for
Scenaria’s industry expertise and analyt-
ics capability to make an impact.
For my part, I feel fortunate to be able
to engage personally and professionally
with these folks (I was taught long ago to
surround myself with likable people who
are different from me and smarter than
I am). In particular, I am thankful to IN-
FORMS for creating the BAOR confer-
ence and thus establishing a compelling
space for hatching these types of new
connections. Moreover, meeting people
like Scott and Sandy has also convinced
me that there are many different types of
analytic leaders out there – and I can’t
wait to see more of them at the next spring
party (April 7-9, 2013 in San Antonio).
We all have an awful lot to learn. ❙
Vijay Mehrotra ([email protected]) is an
associate professor, Department of Finance and
Quantitative Analytics, School of Business and
Professional Studies, University of San Francisco.
He is also an experienced analytics consultant
and entrepreneur and an angel investor in several
successful analytics companies. He is a senior
INFORMS member.
analy z e t h i s !
General Chair
Ronald G. Askin
Arizona State University
Program Co-Chairs
John W. Fowler
Esma Gel
Arizona State University
INFORMS ANNUAL MEETING
PHOENIX 2012
Register Today!
INFORMATICS
RISING
INFORMATICS
RISING
October 14-17, 2012
Phoenix Convention Center, Phoenix, AZ
Plus
INFORMS-SMDM Joint Program
in conjunction with the
Society for Medical Decision Making
Special Joint Program
Wednesday, October 17
Followed by the
SMDM Annual Meeting
October 17-20
Thanks to Our Sponsors
http://meetings2.informs.org/Phoenix2012
Phoenix 5th color ad_Full page Ad 5/9/12 4:35 PM Page 1
NOTES & FURTHER READING
1. For the interested reader, here’s a link to a story
about my frst fall conference: www.orms-today.
org/orms-12-03/frsomething.html .
2. A video of Scott’s 2011 presentation is available
online at www.informs.org/Community/Analytics/
Videos/Analytics-Process-Presentations .
3. An abstract of Sandy’s 2012 presentation is
available online at http://meetings2.informs.org/
Analytics2012/analyticssustainability.html .
www. i nf or ms . or g
We all love the “hard” side of analytics. What
we often struggle with as analysts is what you
might call the “soft” side of analytics, which is
always more challenging than the “hard” stuff.
Here are a few of the reasons why.
Many times, the problem is not insuffcient
data, defective data, inadequate data models or
even incompetent analysis. Often, the reason
that better decisions are not made in less time
is that many companies of all sizes have some,
if not many, managers and leaders who struggle
to make decisions with facts and evidence . . .
even when it is spoon-fed to them. One reason
is that regardless of functional or organizational
orientation, some executives tend not to be ana-
lytically competent or even interested in analy-
sis. As a result, they tend to mistrust any and all
data and analyses, regardless of source.
In other situations, organizations still discount
robust analysis because the resulting implica-
tions require decisions that confict or contrast
with “tribal knowledge,” institutional customs,
their previous decisions or ideas that they or
their management have stated for the record.
Something to keep in mind is that at least some
of the analysis may need to support the current
thinking and direction of the audience that is an-
alytically supportable if you want the audience to
listen to the part of your analysis that challenges
current thinking and direction.
Understanding the context or the “why?” of
analysis is fundamental to benefting from it.
However, there are times when the results of an
analysis can be conficting or ambiguous. When
the results of analysis don’t lead to a clear, un-
arguable conclusion, then managers or execu-
tives without the patience to ask and understand
“why?” may assume that the data is bad or, more
commonly, that the analyst is incompetent.
Perhaps the most diffcult challenge an or-
ganization must overcome in order to raise the
level of its analytical capability is the natural hu-
bris of senior managers who believe that their
organizational rank defnes their level of unaided
analytical insight. Hopefully, as we grow older,
we also grow wiser. The wiser we are, the slower
we are to conclude and the quicker we are to
learn. The same ought to be true for us as we
progress up the ranks of our organization, but
sometimes it isn’t.
If these are the reasons for the organiza-
tional malady of failing to fully leverage ana-
lytics to make higher quality decisions in less
time, what is the remedy?
For the analyst, I recommend the following:
1. Put yourself in the shoes of the decision-
maker. Try to step back from the details
of your analysis for a moment and ask
yourself the questions he or she will ask.
2. Engage your decision-maker in the
process. Gather their perspective as an
input. Don’t make any assumptions. Ask lots
of questions. They probably know things that
you don’t know about the question you are
trying to answer. Draw them out. Schedule
updates with the decision-maker, but keep
them brief and focused on essentials. Ask
for their insight and guidance. It may prove
more valuable than you think.
3. Take time to know, explore and
communicate the “why?” of your analysis.
Why is the analysis important? Why are
the results the way they are? To what
factors are the results most sensitive and
why? Why are the results not 100 percent
conclusive? What are the risks and why do
they exist? What are the options?
4. Make sure you schedule time to
explain your approach and the “why?”
Your decision-maker needs to know
beforehand that this is what you are
planning to do. You will need to put the
“why”? in the context of the goals and
concerns of your decision-maker.
5. Consider the possible incentives
for your decision-maker to ignore
your recommendations and give him
or her reasons to act on your
recommendations that are also consistent
with their own interest.
6. “A picture is worth a thousand words.”
Make the analysis visual, even interactive,
if possible.
7. Consider delivering the results in Excel
(leveraging Visual Basic, for example), not
10 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
why the soft side of
analytics is so hard to
manage
By aRnOlD maRk wells
however, there are times
when the results of an
analysis can be conflicting
or ambiguous. when the
results of analysis don’t
lead to a clear, unarguable
conclusion, then managers
or executives without
the patience to ask and
understand “why?” may
assume that the data is
bad or, more commonly,
that the analyst is
incompetent.
f OR u m
11 | a na ly t i cs - maga z i ne . or g
just in a Power Point presentation or
a Word document. In the hands of
a skilled programmer and analyst,
amazing analysis and pictures can
be developed and displayed through
Visual Basic and Excel. Every
executive already has a license for
Excel and this puts him or her face-
to-face with the data (hopefully in
graphical form as well as tabular).
You may be required to create a
Power Point presentation, but keep
it minimal and try to complement
it with Excel or another tool that
actually contains the data and the
results of your analysis.
Frustration with your decision-mak-
ing audience will not help them, you
or the organization. Addressing them
where they are by intelligently and
carefully managing the “soft” side of
analytics will often determine whether
you make a difference or contribute to
a pile of wasted analytical effort. ❙
Mark Wells ([email protected]) is a
principal with End-to-End Analytics. For 20 years, he
has consulted with the management of global supply
chains in the chemicals, medical devices, consumer
goods, high tech, automotive and retail industries, as
well as the public sector. Wells holds an MBA (focusing
in operations management and operations research)
from Drexel University. He publishes a supply chain
action blog on supply chain and analytics, from which
this article is adapted. He is an INFORMS member.
f OR u m
INFORMS CREDENTIALING
TASK FORCE
• Terry Harrison (Penn State)
• Lisa Kart (Gartner)
• Bill Klimack (Chevron)
• Dave Leonhardi (Boeing)
• Jack Levis (UPS)
• Polly Mitchell-Guthrie (SAS)
• Scott Nestler (Naval Postgraduate School)
• Michael Rappa (NC State/Inst. Adv. Analytics)
• Anne Robinson (Verizon Wireless)
COMING APRIL 2013...
SPONSORED BY INFORMS
ANALYTICS CERTIFICATION

Members interested in becoming a Certified Analytics Professional (CAP)
make plans now to take the profession's first general analytics certification
exam. Offered at 2013 INFORMS Conference on Business Analytics and
Operations Research, April 7-9, 2013, San Antonio, Texas.
Watch future issues
of OR/MS Today and
Analytics Magazine
for more details.
Read “The Shape
of Analytics
Certification” in
February 2012 issue
of OR/MS Today.
BENEFITS OF CERTIFICATION
• Advances your career potential by setting you
apart from the competition
• Drives personal satisfaction of accomplishing a
key career milestone
• Helps improve your overall job performance
by stressing continuing professional development
• Recognizes that you have invested in your
analytics career
• Boosts your salary potential by being viewed as
experienced analytics professional
ELIGIBILITY
• BA/BS or MA/MS degree in related area
• At least five years of analytics work-related
experience for BA/BS holder
• At least three years of analytics work-related
experience for MA/MS holder
• Verification of soft skills/provision of
business value by employer
frustration with your
decision-making audience
will not help them, you
or the organization.
addressing them where
they are by intelligently
and carefully managing
the “soft” side of analytics
will often determine
whether you make a
difference or contribute to
a pile of wasted analytical
effort.
subscri be t o Anal yt i cs
It’s fast, it’s easy and it’s FREE!
Just visit: http://analytics.informs.org/
www. i nf or ms . or g
Competition in analytics is a
familiar concept on an orga-
nizational level. An “analyt-
ics gap” exists between the
predictive haves and have-nots: an insur-
ance company that accurately predicts a
new customer’s actuarial risk, a mortgage
lender that better predicts the probability
of default, a retailer that better predicts
its churn rate, a social network its num-
ber of followers or an ad platform its click-
through rates. If any of these companies
can accomplish their predictions better
than the rest of their industry, they gain a
distinct competitive advantage – survival
of the analytic fttest.
What happens when this same ap-
proach is applied to the building of the
predictive models themselves? The rise
of “competitive research” has shown that
as much or more value can be derived
from competition between the predic-
tive analysts. In this article, we’ll exam-
ine the past results of making analytics
By maRgit zwemeR
C
p R e D i ct i ve mODe l i ng
12 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
crowdsourcing the full analytics value chain.
Research as a
competitive sport
The rise of “competitive research” has shown that as much or more value can be derived
from competition between predictive analysts.
www. i nf or ms . or g 13 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
a competitive sport and see how this ap-
proach is affecting the full analytics value
chain, from problem identifcation to anal-
ysis to implementation, and even how
analysts might be recruited and compen-
sated in the near future.
ThE hISTOry OF INNOvATION prIzES
We tend to think of crowdsourcing as a
cost-effective way to accomplish low skill,
repetitive tasks, but a quick glance at the
history of innovation prizes shows that open
competition can achieve breakthrough re-
sults in high value-added areas. The tra-
ditional methods of allocating research
funding and corporate analytics budgets
can encourage a risk averse and cautious
mentality . . . and solutions that are “good
enough” but not great. The winner-take-all
nature of prizes breeds a form of intelligent
risk-taking that is particularly good at solv-
ing previously intractable problems. Essen-
tially, it is a way of crowdsourcing genius.
As far back as the 18th century, the
British government offered more than
£100,000 in prize money to anyone who
could come up with simple and practical
methods for measuring longitude to assist
maritime navigation. A watchmaker, John
Harrison, won the task. Then there was
the very frst non-stop transatlantic fight in
1927, which awarded Charles Lindberg, a
relatively unknown aviator at the time, the
$25,000 Orteig Prize. In 2004, a privately
funded team led by engineer Burt Rutan
captured the $10 million Ansari X Prize
for becoming the frst non-governmental
organization to launch a man into space
not once but twice within two weeks using
a largely reusable spacecraft. The 2009
Netfix Prize demonstrated that this model
can be applied equally well to algorithmic
innovation. Netfix offered a million dollars
for a 10 percent improvement in its movie
recommendation algorithm. An estimated
80 percent of the movies that customers
watch on Netfix are found through the
recommendation engine, so, given the
number of Netfix customers, the million-
dollar prize was money well spent.
All of the prize contests mentioned
here produced remarkable accomplish-
ments, and while they could have or
would have been achieved without such
bounties to spur them on, it is unlikely
they would have happened so quickly.
MODErN INNOvATION prIzES
Using innovation prizes to not only
solve a one-off problem but as a regular
part of doing business is catching on in
many different areas. Topcoder is a well-
known community of more than 400,000
developers who compete in challenges
ranging from architecting an entire soft-
ware system to creating a Web site to
developing new algorithms. The commu-
nity competes to design and then build
each module of the product, with a prize
linked to each component based upon its
diffculty. Innocentive is another innova-
tion platform that hosts an even wider ar-
ray of challenges, from fnding biomarkers
for ALS (Lou Gehrig’s disease) to invent-
ing a low-cost rainwater storage system.
From brainstorming, to theory, to blue-
prints, to code, all of these can now be
crowdsourced.
Before I dive more deeply into re-
search as a competitive sport, I should
tell you that I work for Kaggle, a predic-
tive modeling competition platform. If the
examples I bring up seem to tilt heavily
toward my own organization, it’s because
this is the area of competitive research I
know best, and for which I can marshal
hard evidence to back my claims.
Inspired by the Netflix Prize, Kaggle
was founded in 2010 to apply the innova-
tion prize model and digitize it, creating
cR OwD s Ou R ci ng
For many years, INFORMS has offered a Job Placement Service to connect
employers and qualified O.R. and analytics professionals. This service can be
used alone or in conjunction with the Annual Meeting Job Fair at the 2012 Annual
Meeting. Both give applicants and employers a convenient venue for meeting. JPS
is free to INFORMS member applicants.
http://jps.informs.org
LOOKING?
HIRING?
For info on job fair at the 2012 INFORMS Annual Meeting,
visit http://meetings.informs.org/phoenix2012/jps.html
www. i nf or ms . or g 14 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
a marketplace for data science. Kaggle
hosts prediction competitions that solve
large-scale data problems in areas such
as business, health, education and sci-
ence. The competitions lead to more
accurate algorithms for the companies
that sponsor them because they pit a
wide range of solutions and techniques
against each other on a data science
“proving ground.” The online communi-
ty has grown to more than 40,000 data
scientists and predictive analysts, com-
peting under the slogan “making data
science a sport.”
While the painstaking work of building
a great predictive model could not seem
further from the mud and sweat tradition-
ally associated with sports, the competi-
tive dynamic between Kaggle participants
is what has consistently produced models
that are not just better than the current
state of the art, but better than even the
analysts involved thought possible.
Kaggle competitions have dramati-
cally outperformed pre-existing bench-
marks in every competition run. For
example, the best performing algorithm
in Allstate’s Kaggle competition pro-
duced a 271 percent improvement over
the internal version and an 83 percent
improvement over the expected best-
case development of a similar model
using internal or traditional third-party
resources. Allstate has validated that
the Kaggle competition produced a
model that performed substantially bet-
ter than what they would have expected
to develop internally or with a special-
ized outside predictive modeling con-
sulting firm. In a competition for another
client, they allowed their internal ana-
lytics team to compete anonymously in
the competition. While they were not
the overall winners, their performance
well exceeded the benchmark that they
themselves had built before the compe-
tition started.
The Kaggle platform also includes a
real-time leaderboard associated with
each competition. Getting instant feed-
back on the accuracy of their models
against a hidden validation set, rather than
submitting models that are only judged at
the end of the competition, encourages
contest participants to push themselves
harder in order to leapfrog their competi-
tors. The competitive dynamic drives data
scientists to continue exploring ways to
improve predictive accuracy, spurred on
by the knowledge that someone else has
found something that makes their solu-
tion better.
Figure 1 shows the change in leader
and increase in predictive accuracy in the
Kaggle-NASA mapping dark matter com-
petition. Within a week of the competition
launch, the benchmark based on more
than 10 years of physics research was
beaten by … a glaciologist. The same
asymptotic pattern occurs in all Kaggle
competitions and often with the same
surprising winners.
Participants are located all over the
world and often work in felds that are, on
the surface, completely unrelated to the
source of the data problem, such as the
hedge fund trader who recently won an
education technology challenge or the
cR OwD s Ou R ci ng
Figure 1: Predictive accuracy of the Kaggle-NASA
mapping dark matter competition over time.
Figure 2: The universe of Kaggle participants offers
a diverse skill set.
while the painstaking
work of building a great
predictive model could not
seem further from the mud
and sweat traditionally
associated with sports,
the competitive dynamic
between participants is
what has consistently
produced models that are
not just better than the
current state of the art,
but better than even the
analysts involved thought
possible.
cR OwD s Ou R ci ng
www. i nf or ms . or g 15 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
neuroscience student solving an air pol-
lution problem (see Figure 2). Kaggle’s
focus on data competitions with real-time
leaderboard feedback brings out fresh tal-
ent and drives objectively better results.
prEDICTINg ThE FuTurE OF
prEDICTINg ThE FuTurE
Using competitions to crowdsource
analytic talent does not mean that in-
house analytics groups become obso-
lete, but it does change how they will
approach their role. In the future, the
most valuable skill may not be solving
the analytics problem itself, but being
able to clearly define and structure the
problem into a competition format and
evaluate the results to determine which
model can be most successfully applied
to your particular business. By introduc-
ing open, competitive research into the
analyst’s toolset, an opportunity arises
to leverage a large analytics talent pool
in a cost-effective, scalable way.
Kaggle’s own expansion shows how
competitive research can be applied to
many other points along the analytics
value chain. In the past year, Kaggle has
introduced specially formatted, invita-
tion-only competitions in which a select
group of past competition winners are
invited under non-disclosure agreement
to work on sensitive datasets. Another
recently introduced type of competition,
marketed as Kaggle Prospect, extends
the crowdsourcing model to the prob-
lem identification phase that happens
before the competition can be defined.
Hosts share a sample of their raw data
with Kaggle’s participants, who can use
the data to suggest predictive models
that could be built, uncovering new
or less obvious questions that further
analysis of the data can explore. At the
other end of the analytics value chain,
Kaggle makes it simple to operational-
ize the best predictive models from the
competitions and integrate them into
existing systems by hosting and imple-
menting the model and making it acces-
sible to the client through an application
programming interface.
Competitive research has also
created a reputation engine for the
analytics industry. Recently, Kaggle in-
troduced recruiting competitions, which
allow companies to filter their data sci-
ence applicants on demonstrable skills
rather than resumes. Instead of leafing
through a stack of resumes and then
using 10-minute technical brain teasers
to determine if an applicant is skilled at
solving large, open-ended data prob-
lems, a recruiting competition allows
organizations to “try before they buy.”
The first competition of this type was
launched for Facebook and is one of the
most popular contests that Kaggle has
hosted. Competitive research, whether
based on past competition results or
a recruiting competition for a specific
company, in an objective and unbiased
way to filter talent. Screening applica-
tions through a data science competi-
tion decreases both the false positive
and false negative rates – no more bad
interviews and no more having great
candidates waiting forever for the phone
to ring because they come from a non-
traditional background.
Another implication of competitive
research is in how analysts are com-
pensated. A hedge fund trader who is
great at predicting risk makes millions,
so why is an equally smart analyst in a
different industry, who is also great at
predicting risk, not compensated nearly
as well? In most industries, predictive
modeling is still treated as a cost center,
even if the results of the model have a
direct impact on the company’s profit-
ability. The demand for people with the
skills to crunch large amounts of data
far exceeds the supply, but salaries are
much more sticky. The creation of com-
petitive marketplaces for analytics is
seeing the size of the prize pools start
to converge toward the business value
of the models as the hosts bid up prize
money to attract the best talent to their
problem.
CONCLuSION
Selecting the best predictive model
through a research competition is, as
Sloan School Professor Andy McAfee
so eloquently describes it, “not picking
a horse, but hosting a horse race.” Data
research competitions are a resource-
effcient way for organizations to solve
complex data problems, and they create a
meritocratic market for talent that chang-
es the way analysts work. As the crowd-
sourced model for competitive analytics
catches on, companies have less need
to build large, in-house analytics teams,
but the people they do hire need to un-
derstand how to structure their questions
for competitive research. May the best
model win. ❙
Margit Zwemer ([email protected]) is a data
scientist and community manager at Kaggle.
Request a no-obligation infORms member Benefits packet
For more information, visit: http://www.informs.org/Membership
www. i nf or ms . or g
collaborative forecasting enables companies to transition from periodic, disparate and
isolated forecasting activities to a single, real-time enterprise forecasting process.
Forecasting is the foundation
of virtually all business plan-
ning. Scheduling production
requires a demand forecast.
Deciding between short- and long-term
investments requires a cash fow forecast.
You need a customer usage forecast in
order to buy enough servers to support
a growing cloud-based software applica-
tion. The list of forecasts that a company
must prepare is endless and, moreover,
the forecasts are intertwined.
One of the biggest problems when
forecasts are inextricably linked is the lack
of communication between stakeholders.
No one knows which assumptions have
been incorporated, whether they are up
to date or which data is relevant. The re-
sult is incomplete information and poten-
tially obsolete forecasts. This uncertainty
leads to confusion and mistrust in the
forecast. Stakeholders start looking for
alternative forecasts they can confdently
base key decisions on; sometimes this
means building their own models, spread-
sheets or local data systems. Other times
it means relying on their experience and
gut-feel alone. While they may feel more
confdent, they still have no assurance
that data and assumptions are correct
and up to date.
Because forecasting is so pervasive
and so important, it should be treated as
the “plumbing application” that it truly is.
Imagine if all forecasting activities were
integrated into a single, living enterprise
forecast. Data is automatically analyzed
as soon as it becomes available. People
with relevant insight about the business
and knowledge about planning decisions
input this information as soon as it chang-
es. The impacts of changes to data and
assumptions automatically fow through
the system in real time, updating all rel-
evant forecasts. You can see the impacts
on all areas of the organization as chang-
es happen. You can test and understand
the implications of what-if scenarios and
collaborative forecasting:
from vision to reality
By BRian lewis
F
B u s i ne s s p l anni ng
16 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
17 | a na ly t i cs - maga z i ne . or g
cOl l aBOR at i ve f OR e cas t i ng
A leading marketing company’s scoring models used to take
4.5 hours to process. Now, with high-performance analytics
from SAS,
®
they’re scored in 60 seconds.
Get the relevant insights you need to make decisions in an
ever-shrinking window of opportunity – and capitalize on
the complexity of big data to differentiate and innovate.
269
What would you do
with an extra 269 minutes?
High-Performance Computing
Grid Computing
In-Database Analytics
In-Memory Analytics
Big Data
high-performance
analytics
A real
game changer.
sas.com/269 for a free white paper
Each SAS customer’s experience is unique. Actual results vary depending on the customer’s individual conditions. SAS does not guarantee results, and nothing herein should be construed as constituting an additional warranty. 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. ©2012 SAS Institute Inc. All rights reserved. S91566US.0412
alternative decisions. You can respond
and modify your plans in real time.
Collaborative forecasting applica-
tions are turning this vision into reality.
These technologies are based on a few
fundamental concepts: balancing ana-
lytics and insight, real-time updates of
data and assumptions, and openness
and transparency.
BALANCINg ANALyTICS AND INSIghT
Every forecasting process falls on a
spectrum that is bounded by two extreme
approaches – pure analytics on one side
and pure insight on the other. Collab-
orative forecasting combines both ap-
proaches, incorporating the best aspects
of each.
Analytics-based forecasting relies
on forecasting “engines” in which sta-
tistical and other mathematical models
are applied to data to automatically fnd
trends, patterns, seasonalities, outliers,
shifts, correlations, etc. Essentially you
push data in and forecasts come out.
These engines employ analytics that
fnd and extract information from data
that humans could not easily fnd them-
selves (if at all). Computing power is dirt
cheap these days, and automating the
analysis of massive amounts of data in
a short period of time is extremely cost
effective.
However, analytics-based forecasts
are only as smart as their underlying
models and algorithms and are highly
dependent on the input data. Forecast-
ing engines cannot fnd information
– such as the effects of a promotion
– unless that information is somehow
represented in the data and the mod-
el knows to look for it. The data itself
might be nonexistent (e.g., no historical
sales for a new product) or irrelevant
(e.g., historical sales do not match a
customer’s new buying pattern). In any
of these cases, analytics-based fore-
casting is not an option.
In contrast, insight-based forecasting
relies exclusively on human judgement
and gut-feel. Key stakeholders apply
their experience, expertise, knowledge
of business operations, expectations of
customer actions, sense of market con-
ditions, etc. to project future outcomes.
This approach has benefts, such as
the incorporation of timely frst-hand
knowledge, but it has faws as well.
People are not particularly adept at
analyzing data for trends, patterns, etc.
and therefore miss out on this contribu-
tion to the forecasts. The approach is
not scalable or easily replicable, which
is a problem if your company needs to
forecast many items. People are also
subject to many sources of conscious
cOl l aBOR at i ve f OR e cas t i ng
www. i nf or ms . or g 18 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
and subconscious biases, such as the
optimism of a sales team, which can in-
appropriately skew forecasts.
Collaborative forecasting software
can balance analytics and insight by
applying analytics to find good base-
line forecasts and then adjusting and
overriding them based on insight. For
each item being forecasted, collabora-
tive forecasting applications must allow
you to select the right mix and applica-
tion of analytics and insight. As anyone
who has worked with large enterprise
software applications can attest, flex-
ibility is a rarity. Fortunately, new col-
laborative forecasting applications are
designed more as forecasting platforms
than forecasting tools and offer users
the high degree of flexibility that they
require.
rEAL-TIME upDATES OF DATA AND
ASSuMpTIONS
As a concept, collaborative forecast-
ing is a “living” process in which fore-
casts are always up to date. As soon as
someone changes an assumption or data
is updated, the forecasts should incorpo-
rate the changes across all departments
in real-time. If the forecasts do not refect
the latest information, then any decisions
you base on those forecasts are immedi-
ately obsolete.
Collaborative forecasting therefore re-
quires a central forecasting application
that pulls in data from disparate systems
spread throughout the organization. The
application must provide mechanisms
for key stakeholders and knowledge-
able people to record their assumptions
and link them directly to the forecasts.
The application cannot simply document
and store insights in some spreadsheet
or database. Instead, assumptions must
be modeled quantitatively and linked to a
forecast. As the assumptions change, the
forecasts change in response.
For example, suppose your marketing
department plans to run a new promotion
that they believe will lift demand by 10
percent next month. Rather than writing a
document or sending an e-mail with these
details, the marketing department should
be able to input key assumptions in the
collaborative forecasting application,
such as the lift percentage, the launch
date and duration, the type of adjustment
(i.e., multiply the lift by the baseline de-
mand), and which forecasts the promo-
tion should adjust. If marketing decides
to delay the promotion by one month, the
demand forecast will automatically re-
fect this shift as soon as they update the
launch date.
Even simple decisions such as de-
laying a promotion can have widespread
operational and financial implications.
The cost to implement the promotion
shifts by one month, which affects finan-
cial forecasts. Did you need additional
headcount to support the promotion? If
so, this affects headcount forecast. The
shifted promotional demand impacts
your revenue forecast and potentially
procurement, inventory and production
forecasts. Remember, all forecasts are
intertwined.
With real-time updates of data and as-
sumptions, collaborative forecasting al-
lows you to see the impacts of changes
as they happen and be more responsive
in planning.
OpENNESS AND TrANSpArENCy
People trust processes that are
transparent and inclusive. Rather than
tracking down and reconciling informa-
tion from Marketing, Sales, Operations,
Finance or HR, let them take direct
ownership of the information they know
best and link to the information that they
need. Collaborative forecasting allows
each stakeholder to add new informa-
tion, propose changes, test scenarios
and share insights in a single, shared
system. Their rationale and contribu-
tions are tracked, time stamped and
documented for complete openness
and transparency.
subscri be t o Anal yt i cs
It’s fast, it’s easy and it’s FREE!
Just visit: http://analytics.informs.org/
with real-time
updates of data
and assumptions,
collaborative
forecasting allows you
to see the impacts of
changes as they happen
and be more responsive
in planning.
19 | a na ly t i cs - maga z i ne . or g
IMpLEMENTINg COLLABOrATIvE
FOrECASTINg
Three potential hurdles exist to im-
plementing a collaborative forecasting
process: technology, awareness and read-
iness. Collaborative forecasting software
now exists in many forms. Some of the
newest and most promising are compre-
hensive and support all of the fundamental
concepts of collaborative forecasting. Oth-
ers are more limited, for example focusing
more on the analytics and the forecasting
engine approach. With new technology
comes new awareness of what is possible
and the interest in collaborative forecast-
ing is increasing rapidly. However, even
with technology and awareness, some
companies may feel they are not quite
ready to implement a collaborative fore-
casting process. You know your process
isn’t great and your forecast accuracy is
only so-so, but process changes can be
challenging at your company. That may
be true, but the fnancial and competitive
advantages of collaborative forecasting
are substantial enough that you cannot
afford to wait any longer. ❙
Brian Lewis, Ph.D. ([email protected]), is
vice president of Professional Services for Vanguard
Software, an enterprise forecasting software company
with more than 2,000 customer in 60 countries,
including 33 of the Fortune 100. He is a senior
INFORMS member.
cOl l aBOR at i ve f OR e cas t i ng
nnnnnnnnnnnnnnnn
nnnnnnnnnnnnnnnn
nnnnnnnnnnnnnnnn
nnnnnnnnnnnnnnnn
nnnnnnnnnnnnnnnn
nnnnnnnnnnnnnnnn
nnnnnnnnnnnnnnnn
nnnnnnnnnnnnnnnn
Decision Sciences Institute
advancing the science and practice of decision making
The Decision Sciences Institute is a nonprofit
professional organization of researchers, managers,
educators, and students interested in decision-
making techniques and processes in private and
public organizations. The Institute is an international
organization with over 3,500 members in 32 coun-
tries. The annual meetings and regional conferences
attract over 3,500 participants a year.
The Decision Sciences Institute
Is Committed to . . .
Research. Afocus on the integration of research
in the art and science of managerial decision making
across traditional functional academic disciplines; an
international forum for presentation and discussion
of research.
Teaching. Aforum for presentation and discussion
of innovative teaching; recognition of teaching
excellence and curriculum innovation.
Practice. An exchange of ideas between leading
professional practitioners and educators.
Benefits Members Receive:
Annual meeting. Members receive discounted fees to
attend, and the meeting draws over 1,500 professionals
together to share current thoughts on theoretical and
applied issues.
Decision Sciences is a highly respected journal among
scholars and is subscribed to by over 1,000 libraries.
It seeks and publishes high quality, theoretical and
empirical articles.
Decision Sciences Journal of Innovative Education is
a high quality, peer-reviewed scholarly journal whose
mission is to publish significant research.
Decision Line, the Institute’s news publication, is
published five times annually and includes feature
columns, as well as information on members, regions,
annual meetings, and placement activities.
Job Placement Services are offered throughout the
year, with position and applicant listings available via
the Internet. Also at the conference, special facilities
aid in position search activities.
www.DecisionSciences.org
Decision Sciences Institute
43rd ANNUAL MEETING
NOVEMBER 17-20, 2012
Join us.
Log on to our website and complete
the membership application to begin
enjoying your benefits.
Regular membership: $160/year
Student membership: $25/year
Documenting the rationale for as-
sumptions is a critical form of commu-
nication between stakeholders. While
there still might disagreement about
the numbers themselves, stakeholders
can have more productive discussions
if they know the rationale behind them.
Perhaps this leads to changes, perhaps
not. But access to information and open
discourse builds trust in the process.
Without it you run the risk of introduc-
ing uncertainty, confusion and mistrust,
and this is something you clearly want
to avoid.
Another reason for documenting ra-
tionale is for performance management.
Everyone knows why they are updating
their assumptions today, but they may
not remember the details in a few days,
let alone a few weeks or months from
now. By time-stamping your changes
and documentation, you have an audit
trail of every change that was made.
When you look back at older forecasts,
you can see the old data and assump-
tions, who was responsible, their ratio-
nale for making changes and so on. By
knowing what was done and whether
or not it was successful, you can de-
sign forecast accuracy and process im-
provements that incorporate the good
and eliminate the bad.
www. i nf or ms . or g
survey of forecasting software reveals interesting trends and new developments.
Time series data is frequently
collected and analyzed, and
then forecasts are made, often
with clever headlines or leads.
These vary from the important to the mun-
dane. Recently, we have seen time series
data and forecasts for unemployment fgures,
gas prices, mortgage loan rates, stock indi-
ces, approval values for President Obama
vs. Mitt Romney, high school science and
math scores, television ratings for “Dancing
With the Stars” and many other examples.
Television and print commentators
frequently show the numbers or graphs,
draw some general conclusion or predic-
tion, but frequently they do not delve into
some underlying issues, such as sample
size, likely voters vs. registered voters,
correlations with other factors, forecasts
are “long-range” and others. Many statisti-
cal and forecasting software products will
automatically make forecasts for time se-
ries data, but did the forecaster know the
specifcs of the model chosen, examine
previous forecasts made on earlier data
or try alternative models? We presume
the answer is “absolutely,” but sometimes
when we read or hear about the forecasts,
we wonder.
For the purposes of this article, I tried
making some forecasts on data that is far
less “weighty” than many of the examples
given above. I collected [1] monthly data
of the total domestic motion picture box-
offce grosses, from January 2000 through
December 2011, or 144 values, and put
these into an Excel spreadsheet. Each
month had just three columns – time peri-
od, date (1/1/2000, 2/1/2000, etc., which
Excel then displayed as Jan-00, Feb-00,
etc.) and the total gross (the numbers are
in millions of dollars and have not been
adjusted for infation). The plan was to
import the spreadsheet into several sta-
tistical or forecasting software products
to make forecasts of the total box-offce
values for January through April 2012,
and then compare those forecasts to the
forecasting an upward trend?
By Jack yuRkiewicz
T
f OR e cas t i ng s Of t waR e s u R ve y
20 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
21 | a na ly t i cs - maga z i ne . or g
actual values for those four months, in
time for the deadline for this article. A
time series plot of the historical data is
shown in Figure 1.
TypES OF FOrECASTINg SOFTwArE
Forecasting software is generally
available from two categorical groups.
The first group, called dedicated soft-
ware, has forecasting capabilities but
does not possess additional statisti-
cal prowess. Thus, dedicated soft-
ware typically can do Box-Jenkins,
exponential smoothing, regression,
nonlinear trend analysis and other
forecasting procedures, but it cannot
find a confidence interval for the pop-
ulation proportion or do a factor analy-
sis. An example of dedicated software
is Forecast Pro. The second group
consists of general statistical analy-
sis products that include forecasting
capabilities. Some examples are IBM
SPSS Statistics, SAS, Minitab, Stat-
graphics, NCSS and Systat. One pos-
sible advantage of dedicated software
is that it may have certain procedures
or capabilities (e.g., ARIMA interven-
tion, econometric, transfer function
models, etc.) that the general statisti-
cal products might not.
Another attribute of forecasting
software is its level of automation;
that is, the degree to which the soft-
ware can specify the appropriate fore-
casting model to use on your data.
There are three levels of automation.
The first may be called automatic. Au-
tomatic software advises the “appro-
priate” model for the particular data
set. That is, it recommends a fore-
casting model or procedure by mini-
mizing some statistic (e.g., Akaike
Information Criterion (AIC), Schwarz
f OR e cas t i ng s Of t waR e s u R ve y
Call for Applications
SPONSORED BY INFORMS SECTION ON ANALYTICS
The purpose of the Innovation in Analytics Award is to recognize creative and unique
developments, applications or combinations of analytical techniques. The prize is not meant
to recognize strictly theoretical advances, though theoretical advances might be an enabler
to innovative applications. Similarly, the prize is not focused on implementation value
created, but such value creation might add credibility to the innovation.
Applicants must submit a brief (500-1,000 word) summary of their work by July 15, 2012.
2011-2012 COMMITTEE CHAIR
Michael F. Gorman
University of Dayton
Voice: +1-937-229-3382
E-mail: [email protected]
Key Dates
Deadline for applications – July 15, 2012
Semi-finalists notified – July 31, 2012
Semi-finalists presentations – October 15-16, 2012, Phoenix, AZ at
INFORMS Annual Meeting, Innovations in Analytics Track
Finalists selected by – November 30, 2012
Finalists' presentations – April 7-8, 2013, San Antonio, TX at
INFORMS Conference on Business Analytics and Operations Research
Judging Criteria
• Level of uniqueness, creativity, value and (potential) contribution of the analytical
techniques developed
• Innovation in any of the three dimensions of analytics: descriptive, predictive or
prescriptive. Submissions that span more than one dimension of analytics are preferred
• Implemented approaches are preferred but not required; to the extent implemented
savings are claimed in the submission, the judging panel will seek verification of the
implementation
• Submissions that have the potential for more widespread use will be given added
consideration, but particularly clever innovations with narrow focus are also highly valued
ADDITIONAL INFORMATION
For more information on application process, eligibility, and prize amounts, see
http://www.informs.org/Community/Analytics/News-Events.
SECTION ON
ANALYTICS
Innovation in Analytics Award
2
nd
Annual
Figure 1: Total monthly domestic motion picture box-offce gross sales 2000-2011.
www. i nf or ms . or g 22 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
Bayesian Information Criterion (SBIC),
mean square error (MSE), etc.) and
then finds the optimal parameters for the
model, calculates forecasts and confi-
dence intervals, gives various summary
statistics and makes graphs. The user
can override the recommended proce-
dure, specify some other forecasting
technique, and the software then finds
the optimal parameters for that model,
gets the forecasts, etc. Most dedicated
forecasting products can operate in the
automatic mode; fewer general statisti-
cal products (e.g., IBM SPSS Statistics
and Statgraphics) can be put into this
category.
The next automation level is semi-
automatic. Here, the user specifies the
particular forecasting methodology the
software should use, for example, a
Box-Jenkins model, and the software
proceeds to find the optimal parameters
of that model, the forecasts, summary
statistics, graphs, etc. Most general
statistical products operate in the semi-
automatic mode.
Finally, the third level of automation,
dubbed manual, requires the user to
specify both the model and the param-
eters of that model. Thus, for manual
software, if the user specifies Winters’
multiplicative model, he or she must
also enter the three smoothing con-
stants. The software will then give the
forecasts, plots, summary statistics,
etc. After examining the summary sta-
tistics, forecasts, etc., the user manu-
ally enters new model parameters and
repeats the process. Finding the “op-
timal” model parameters can thus be-
come a tedious trial-and-error process.
The standard advice still holds: If you
frequently forecasting time series data,
you should consider using an automatic
or semi-automatic product.
f OR e cas t i ng s Of t waR e s u R ve y
wOrkINg wITh ThE SOFTwArE
This article is not meant to formally re-
view any product, but I used the month-
ly box-offce data in the latest versions
of SPSS, Forecast Pro, Statgraphics,
NCSS, Minitab and Systat, all of which I
have used for some time. These products
represent a cross-section of the three
automation categories described. All
the products read and imported my Ex-
cel spreadsheet, usually but not always,
without additional tweaking (Forecast
Pro needs, in separate cells, information
about time span, seasonality, etc.).
All the programs I tried were easy to
use. Menu systems are clear (I adhered
to the time honored adage of not click-
ing on Help unless and until it was a last
resort) and the output was easy to read
and interpret. If the software had an au-
tomatic mode, I always used that frst.
Looking at the monthly box-offce data in
Figure 1, repetitive “peaks” and “valleys”
appeared over the years. The data shows
high box-offce values from May through
July, and secondary peaks in November
and December. Low box-offce months
are typically in February and September.
These rules, off course, do not always
apply (e.g., December 2009), but there
appears to be monthly seasonality and
an upward trend to the numbers. Thus,
a Box-Jenkins or a Winters’ exponential
Figure 2: Forecast Pro’s recommendation.
Figure 3: Overriding Forecast Pro’s recommendation and specify-
ing a Box-Jenkins model instead.
smoothing model could be two method-
ologies to apply to this data. Forecast Pro
recommended Winters’ method with ad-
ditive seasonality (see Figure 2). When
I overruled that advice and asked for a
Box-Jenkins model, it recommended an
ARIMA(1,0,1)x(0,1,2) model (see Figure
3). IBM SPSS Statisitics recommended
an ARIMA(2,0,12) model. The third auto-
matic product, Statgraphics, after I told it
to analyze all the models available and
fnd the one with the smallest AIC, had
its StatAdvisor recommend an ARIMA
(0,1,1)x(2,1,2)12 model in which “a mul-
tiplicative seasonal adjustment was ap-
plied” (see Figure 4).
www. i nf or ms . or g 23 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
FOrECAST rESuLTS AND A CAvEAT
As mentioned in previous forecasting
surveys, a particular data set could result
in automatic or semi-automatic software
recommending different models and/or
different parameters for those models.
Figure 5 shows the box-offce forecasts
obtained from some of these products.
Because SPSS and Statgraphics both
recommended Box-Jenkins models, I told
Forecast Pro to utilize its recommended
procedure, an exponential smoothing
model, and then fnd the appropriate Box-
Jenkins model and make forecasts for
both. As NCSS does not make a model
recommendation, I specifed it should use
the Box-Jenkins procedure and fnd the
appropriate parameters. The last column
shows the actual box-offce totals for the
frst four months of 2012. For March 2012,
all the products gave forecasts that were
much lower than the actual box-offce fg-
ure. Perhaps “The Hunger Games,” re-
leased on March 23 and which grossed
$233 million in just those nine days of
March, may have something to do with
those low projections.
On May 15, ABC News reported, “Today,
just over a third of U.S. adults are obese. By
2030, 42 percent will be, says a forecast re-
leased Monday. That’s not nearly as many
as experts had predicted before the once-
rapid rises in obesity rates began leveling
off” [2]. The article intimated that the Cen-
ters for Disease Control and Prevention had
made the forecast. Thus, we are getting a
forecast of the fraction of Americans who
will be considered obese 18 years from
now, and that’s a lower fraction that was
forecast who knows when. In my box-offce
example, all I wanted to do is to make box-
offce forecasts four months into the future,
and our forecasting software substantially
underestimated the box-offce grosses that
would occur in month three. Perhaps when
we read about such long-range (and even
short range) forecasts, or contemplate
making them ourselves, we might pause
and think “The Hunger Games.”
ThE SurvEy
For this year’s forecasting software
survey, as in the past, we tried to identify
as many forecasting products as possible.
We e-mailed the vendors and asked them
to respond to our online questionnaire so
readers could see the features and ca-
pabilities of the software. The purpose of
the survey is to inform the reader of what
is available. The information comes from
the vendors, and no attempt was made to
verify the information they gave us. Inevi-
tably, after the results are published, we
hear, “How could they have left out (my)
product X!” Thus, if we did make an omis-
sion, we ask you to please accept our
f OR e cas t i ng s Of t waR e s u R ve y
Figure 4: Statgraphics’s StatAdvisor making recommendation for the appropriate models to use. Only the “top 2” choices are shown here.
It also gives the parameters, the forecasts and some specifcs of the calculations for the optimal model (not shown).
Figure 5: Box-offce grosses; forecasts from a sample of automatic and semi-automatic software and the actual grosses, in millions of dollars.
24 | a na ly t i cs - maga z i ne . or g
apology for the oversight. Let us know
of the company and product and we will
add it to online survey results.
If you are interested in getting a
new forecasting program, or just want
to try some product other than the
one you have, you should first look
at the techniques the software offers
and compare those with your needs.
Most, but not all, vendors allow you
to download a time-trial version of
the software, which typically expires
in anywhere from a week to a month.
Make sure the trial version allows you
to work with your own data and check
if any forecasting features or niceties
(typically the data size is one) are
omitted in the trial version. Contact
the vendor with your specific ques-
tions. Users tell us that they found the
vendors to be extremely helpful. ❙
Jack Yurkiewicz ([email protected]) is a professor
of management science in the MBA program at the
Lubin School of Business, Pace University, New York.
He teaches data analysis, management science
and operations management. His current interests
include developing and assessing the effectiveness
of distance-learning courses for these topics. He is a
senior INFORMS member.
f OR e cas t i ng s Of t waR e s u R ve y
1 : an act or the power of
foreseeing : prescience
2 : provident care : prudence
<had the foresight to invest his money wisely>
3 : an act of looking forward; also : a view forward
- Merriam-Webster Dictionary
foresight: n.
Foresight: Te International Journal
of Applied Forecasting is a practical
guide that relates to business forecasting
like no other professional journal can.
Four times each year, Foresight’s pages
are packed with articles, reviews, and
opinions that showcase the best think-
ing and writing in the feld of forecasting.
Every issue features earned expertise fom
practitioners around the globe that will
challenge and inspire you, and that you’ll
incorporate in your day-to-day work,
whatever types of forecasting you do.
Visit forecasters.org/foresight to
learn more and subscribe or renew today.
Foresight.
Do you have it?
Do you want it?
Get it here.
www.forecasters.org/foresight/subscribe
Survey Results & Directory
For the results of the 2012 forecasting software
survey and a directory of forecasting software
vendors, click here.
REFERENCES
1. The authoritative Boxoffce.com website,
followed by many in and out of the motion picture
industry: www.boxoffce.com/statistics/yearly.
2. http://abcnews.go.com/Health/wireStory/20-year-
forecast-shows-end-obesity-epidemic-16295451.
if you are interested in
getting a new forecasting
program or just want to
try some product other
than the one you have, you
should first look at what
techniques the software
can do and compare those
with your needs.
www. i nf or ms . or g
the acceptance of analytics to tackle issues related to big data is going through growing pains, pitting skeptics, who are
typically older and shaped by managing when errors were tolerated, against younger enthusiasts.
A quick quiz: What is a good
nine- or 10-letter description
of the emerging interest in
business analytics and big
data that ends in “-al”?
A choice that may come to mind for
many is “hysterical.” This choice reflects
frenzied excitement about opportunities
for business analytics to solve problems
often resulting from big data. Advocates
– actually enthusiasts – of analytics have
become energized by the growing inter-
est in the fields of business intelligence
and data mining. But perhaps a less ob-
vious choice of “skeptical” is an equally
valid answer. Doubters and naysayers
of business analytics believe the inter-
est in these topics is overblown and
misguided.
whErE ArE ThE SkEpTICS COMINg
FrOM?
Let’s start with the view of the skeptics
of analytics. Who are they? What is their
profle? What is their objection to embrac-
ing and deploying analytics?
analytics & big data:
skeptics vs. enthusiasts
By gaRy cOkins
A
D R i v i ng ch ange
25 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
www. i nf or ms . or g 26 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
I will probably exaggerate, but here is my
take on the skeptics of analytics. They are
likely to be over the age of 50. When they
took a statistics course in college, many of
them probably just wanted to pass with a
“D” and get the course behind them.
When the skeptics’ careers were shaped,
they did not have PlayStations, Nintendos
or Xboxes – no video games at all. They
did not have 150 channels on cable, satel-
lite TV, video movies or DVDs. They had no
surround-sound, no smartphones, no per-
sonal computers and no Internet.
During the skeptics’ early careers, they
observed and participated in company fre-
fghting. This was actually fun for them. It
kept them busy. The assertive skeptics were
continuously promoted to higher job posi-
tions because of their quick wit and intuition.
These types of managers are not self-serving
sycophants and political players (although
some are). They are hard workers.
The older skeptics grew up watching
television series like “Ozzie and Harriet”
and “Leave It to Beaver.” At their offces
– since most skeptics have been white-
collar workers most of their careers –
their co-workers and supervisors weren’t
much different from Ozzie Nelson and
Ward Cleaver, the fathers in those TV se-
ries. Ozzie and Ward shaped the skep-
tics’ values and attitudes of what the daily
workday would be like. After a pleasant
breakfast at home with the family, one
headed to the offce to shuffe papers, at-
tend meetings and return missed phone
calls (from handwritten secretary notes).
The career experiences of skeptics did
not involve punctuated change and vola-
tility as occurs in today’s sped-up world.
However, it was not easy for them. Their
organizations weren’t static and frozen in
time. Occasional new products and ser-
vices were developed, and new types of
customers pursued. Some companies
merged, divested and acquired other
companies. Periods of transition followed
those events before the skeptics’ work-
weeks settled down to a normal pace. But
just like after a passing storm, they even-
tually returned to shuffing papers, writing
memos and attending meetings.
Skeptics were not free from solving
problems or evaluating opportunities.
Those are eternal tasks in anyone’s work-
day. What was different about the solu-
tions developed when skeptics’ careers
were shaped is the solutions just weren’t
that elegant – they were like Swanson TV
dinners, oven-baked in aluminum trays.
rELIANCE ON BuFFErS TO prOTECT
AgAINST ErrOrS
Here is an example of problem-solv-
ing in manufacturing that the skeptics
experienced. If a supplier’s shipment of
component parts was going to be late,
that was OK. The assembling manufac-
turer simply rescheduled their customers’
orders, and its shipment would likely be
late to the distributor or retail store. If this
resulted in an out-of-stock shortage when
customers wanted or needed the fnished
product, then their customer just dealt
with it and made do. The reasoning was
that sometimes it rains, and you don’t
have an umbrella. And Murphy’s Law of
the unexpected periodically sneaks up.
When skeptics’ careers were shaped,
there were no crystal balls to predict the
future. But they did not need them. In
manufacturing a forecast was somehow
produced every month or so, and it was
used to establish plans for determining
what and how much inputs to buy (e.g.,
raw materials, component parts) and what
types of resources to hire or purchase.
When skeptics’ careers were shaped,
they survived using buffers to protect
them from errors and missed delivery
schedules. Buffers were the magic elixir
that kept problems from becoming larger
or more painful. Skeptics relied on three
types of buffers related to time, fow and
resources. They would start things ear-
lier to buffer expected fnish dates, build
and stock extra inventory to buffer mate-
rial shortages, and add more people and
equipment to buffer capacity.
analy t i cs & B i g Data
when skeptics’ careers
were shaped, they
survived using buffers to
protect them from errors
and missed delivery
schedules. Buffers were
the magic elixir that kept
problems from becoming
larger or more painful.
hel p promote Anal yt i cs
It’s fast and it’s easy! Visit:
http://analytics.informs.org/button.html
27 | a na ly t i cs - maga z i ne . or g
Let’s fast-forward to today. Skeptics
recognize that the world has changed.
It is much more volatile. But they still
challenge the need to embrace deep or
advanced analytics. When their careers
were shaped they observed people
measuring something with a microm-
eter, marking it with a piece of chalk
and cutting it with a swinging ax. They
behaved this way too. Why be precise?
Few organizations recognized the pen-
alties and extra buffer-related expenses
and investments that mitigated against
imprecision and errors. Ineffciencies,
long delivery lead times and temporary
shortages were tolerated by both sup-
pliers and their customers.
whErE ArE ThE ENThuSIASTS OF
ANALyTICS COMINg FrOM?
Let’s look next at the opposite view
– that of the analytical enthusiasts
whose careers have been more re-
cently shaped. Who are they? What is
their profle? What is their enthusiasm
– sometimes hysterically so – for em-
bracing and deploying analytics?
Again I will probably exaggerate to
make my points, but here is my take
on the enthusiasts. They are likely to
be under the age of 40. When they
took their university statistics courses,
they had hand calculators and laptop
computers. (Believe it or not, in my frst
two years in college we only had slide
rules. I graduated in engineering and
operations research in 1971.)
During enthusiasts’ careers, fre-
fghting was not an occasional need – it
was ongoing and never-ending. And its
intensity is not just because there are
more problems (although there are). It
is because there are more opportuni-
ties requiring a sense of urgency.
These younger enthusiasts grew up
watching MTV, “Friends” and “Seinfeld.”
Their sense of humor is more wry and
cynical compared to the skeptics. Their
curiosity about how things work is prob-
ably comparable to the skeptics’ when
they were the same age. However, the
difference is the enthusiasts have far
greater ability to investigate and ana-
lyze – and with more computing power
and more functional software.
In the movie “Moneyball,” Brad Pitt
plays the role of the Oakland Athletics
baseball team general manager Billy
Beane. Just before Pitt fres the team’s
head baseball scout, he says (para-
phrasing), “OK. My turn. When you visit
the homes of an aspiring young baseball
player you tell his parents that he has
a good chance of being a major league
player, you don’t know. You don’t know.”
He repeats that to make his point.
analy t i cs & B i g Data
Save $225
if you are part of a group of 3 or more.
Accommodations at Disney’s Contemporary Resort
are available for $180 per night, per room.
Additionally, specially-priced Theme Park tickets
are available for registrants.
©Disney
Learn more
August 30–31, 2012
Disney’s Contemporary Resort & Convention Center
www.disneyurl.com/DAOS2012
Walt Disney Parks & Resorts is pleased to announce
www. i nf or ms . or g 28 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
This also applies to organizations, es-
pecially where skeptics dominate. Skep-
tics may think they know, but do they
really know?
ENThuSIASTS uSE ANALyTICS TO
rEpLACE BuFFErS
Let’s revisit the example of manufac-
turing and distribution companies. As pre-
viously discussed, in the past the skeptics
solved operational problems with buffers.
But the careers of today’s enthusiasts
have been shaped by buffers that must
be paper-thin. Buffers that the skeptics
enjoyed to protect them from the impact
of problems are too costly today. These
costs and penalties include surplus in-
ventory, excessively long production and
delivery lead times, extra equipment ca-
pacity, poor customer service levels and
more employees than are needed.
Analytics is reducing the size of and
replacing the protective crutch of buffers.
Today, the enthusiasts for analytics are
imaginative and visionary. Their thinking
doesn’t stop with enterprise resource plan-
ning (ERP) systems, which can schedule
part production and purchasing based on
forecast product demand volume, supplier
delivery rates and assembly lead times. En-
thusiasts are far more imaginative than that
– plus they have the analytics and comput-
ing power to be creative.
What enthusiasts do is think forwardly
with probabilistic what-if scenario anal-
ysis. They do not view product distribu-
tion in a supply chain as a linear tree,
branch and leaf structure that sequenc-
es parts like elephants’ trunk-to-tails in
a circus all the way from production to
the customer. Enthusiasts see opportu-
nities in an integrated network of parts
and products that exist or can be made
anywhere. And they then perform itera-
tive trade-off analysis of the interrelated
variables in real time.
Enthusiasts know there are four in-
terrelated variables that have complex
interrelationships: customer service lev-
els, lead times, demand volumes and
unit costs. Since suppliers have mea-
sured and calibrated all of the process-
ing times and consumption rates from
recent past period data, they then know
the relationships among all four vari-
ables. The analytics enthusiasts’ vision is
one where they start with a baseline case
scenario projected into the future. (Most
are already using advanced demand
forecasting techniques.) They then per-
form what-if analysis. They change one
of the variables – probabilistically – as
the independent variable, and calculate
the impact on the other three dependent
variables. And then they change another
variable – and so on.
This gives enthusiasts power to answer
many questions, such as, “What is the ad-
ditional inventory carrying cost if we want
to improve service levels from 97 percent
to 99 percent for our strategic customers?”
These capabilities are no longer a dream or
vision. They exist today.
ENThuSIASTS CAN wIN Buy-IN FrOM
ThE SkEpTICS
Admittedly my profles of skeptics vs. en-
thusiasts are exaggerated. They are not po-
lar opposites but rather people residing along
a continuum. But this does not remove the
challenge of creating a culture for analytics.
My experience is that an effective way
to drive change, overcome resistance and
gain buy-in is through example. Enthusi-
asts can be role models. Lead by example.
Demonstrate what can be done and it will
be done. ❙
Gary Cokins ([email protected]) is an internationally
recognized expert, speaker and author in advanced
cost management and enterprise performance and risk
management systems. He is a principal in business
consulting involved with analytics-based enterprise
performance management solutions with SAS, a global
leader in business intelligence and analytics software.
He began his career in industry with a Fortune 100
company in CFO and operations roles. He then worked
15 years in consulting with Deloitte, KPMG and EDS. His
two most recent books are “Performance Management:
Finding the Missing Pieces to Close the Intelligence Gap”
and “Performance Management: Integrating Strategy
Execution, Methodologies, Risk and Analytics.” He blogs
at http://blogs.sas.com/content/cokins. He is an INFORMS
member.
analy t i cs & B i g Data
today’s enthusiasts have
been shaped by buffers
that must be paper-thin.
Buffers that the skeptics
enjoyed to protect them
from the impact of
problems are too costly
today. these costs and
penalties include surplus
inventory, excessively
long production and
delivery lead times, extra
equipment capacity, poor
customer service levels and
more employees than are
needed.
Join the Analytics Section of INFORMS
For more information, visit:
http://www.informs.org/Community/Analytics/Membership
www. i nf or ms . or g
a statistical measure used to classify time series and infer the level of difficulty in
predicting and choosing an appropriate model for the series at hand.
In 1951 the celebrated Brit-
ish hydrologist H.E. Hurst
published a paper titled, “The
Long-Term Storage Capacity
of Reservoirs.” The paper dealt specifcal-
ly with the modeling of reservoirs, but as
it turned out, the results also held valid for
a number of other natural systems. Hurst
was looking for a way to model the levels
of the river Nile so that architects could
construct an appropriately sized reser-
voir system. While his recommendations
were not implemented (the 1952 Egyptian
revolutions saw to that), he gave life to a
statistical methodology for distinguishing
random from non-random systems and
to identify the persistence of trends, a
methodology known as Rescaled Range
analysis or R/S analysis.
Many years later, while investigating
the fractal nature of fnancial markets –
specifcally, the tendency of a time series
to regress strongly to its mean or to clus-
ter in a direction – noted mathematician
Benoit Mandelbrot happened to stumble
across Hurst’s work and, recognizing the
potential therein, introduced to fractal ge-
ometry, in Hurst’s honor, the term Gen-
eralized Hurst Exponent. Put simply, the
Hurst exponent is used as a measure of
the long-term memory of a time series.
In addition to the Hurst exponent,
Mandelbrot also coined two more terms
useful in describing the long-term mem-
ory of a time series. He called the first
one the Joseph Effect and the second
one the Noah Effect. The Joseph Ef-
fect tells us whether movements in a
time series are part of a long-term trend
and refers to the Old Testament where
Egypt would experience seven years of
rich harvest followed by seven years of
famine. The Noah Effect is the tendency
of a time series to have abrupt changes
and the name is derived from the bib-
lical story of the Great Flood. Both of
these effects in a time series can be in-
ferred from the Hurst exponent.
predictability of time series
By suBiR mansukhani
I
t h e h u R s t e X p One nt
29 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
“Clouds are not spheres,
mountains are not cones,
coastlines are not circles, and bark
is not smooth, nor does lightning
travel in a straight line.”
– Benoit Mandelbrot
www. i nf or ms . or g 30 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
ESTIMATINg ThE hurST ExpONENT
The Hurst exponent is not so much
calculated as it is estimated. A variety of
techniques exist for estimating the Hurst
exponent (H) and the process detailed
here is both simple and highly data in-
tensive. To estimate the Hurst exponent
one must regress the rescaled range on
the time span of observations. To do this,
a time series of full length is divided into
a number of shorter time series and the
rescaled range is calculated for each of
the smaller time series. A minimum length
of eight is usually chosen for the length of
the smallest time series. So, for example,
if a time series has 128 observations it is
divided into:
• two chunks of 64 observations each
• four chunks of 32 observations each
• eight chunks of 16 observations each
• 16 chunks of eight observations each
Steps for estimating the Hurst expo-
nent after breaking the time series into
chunks:
For each chunk of observations,
compute:
• the mean of the time series,
• a mean centered series by subtracting
the mean from the series,
• the cumulative deviation of the series
from the mean by summing up the
mean centered values,
• the Range (R), which is the difference
between the maximum value of
the cumulative deviation and the
minimum value of the cumulative
deviation,
• the standard deviation (S) of the mean
centered values, and
• the rescaled range by dividing the
Range by the standard deviation.
Finally, average the rescaled range
over all the chunks.
The rescaled range and chunk size
follows a power law, and the Hurst ex-
ponent is given by the exponent of this
power law. When the frequency of an
event varies as the power of some quan-
tity associated with the event, it is said
to follow a power law. A wide variety of
natural and manmade phenomena fol-
low a power law. For example, the 80/20
rule (20 percent of the population holds
80 percent of wealth), the winner-take-
all phenomenon, friend connections in a
social network and forest fires all follow
power laws.
INTErprETINg ThE hurST ExpONENT
Using the Hurst exponent we can clas-
sify time series into types and gain some
insight into their dynamics. Here are some
types of time series and the Hurst expo-
nents associated with each of them.
A Brownian time series: In a Brown-
ian time series (also known as a random
walk or a drunkard’s walk) there is no cor-
relation between the observations and a
future observation; being higher or lower
than the current observation are equally
likely. Series of this kind are hard to pre-
dict. Figure 1 provides an example of a
Brownian time series and its estimated
Hurst exponent. The Hurst exponent for
the data plotted above was estimated to
be 0.53 - a Hurst exponent close to 0.5 is
indicative of a Brownian time series.
t i me s e R i e s
An anti-persistent time series: In an
anti-persistent time series (also known as
a mean-reverting series) an increase will
most likely be followed by a decrease or
vice-versa (i.e., values will tend to revert
to a mean). This means that future values
have a tendency to return to a long-term
Figure 1: A Brownian time series (H = 0.53).
simply put, the hurst
exponent is used as a
measure of the long-term
memory of a time series.
Request a no-obligation
infORms member Benefits packet
For more information, visit:
http://www.informs.org/Membership
www. i nf or ms . or g 31 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
mean. Figure 2 provides an example of
an anti-persistent time series and its esti-
mated Hurst exponent.
The Hurst exponent for the data plot-
ted above was estimated to be 0.043.A
Hurst exponent value between 0 and 0.5
is indicative of anti-persistent behavior
and the closer the value is to 0, the stron-
ger is the tendency for the time series to
revert to its long-term means value.
A persistent time series: In a persis-
tent time series an increase in values will
most likely be followed by an increase in
the short term and a decrease in values
will most likely be followed by another
decrease in the short term. Figure 3 pro-
vides an example of a persistent time se-
ries and its estimated Hurst exponent.
The plot shows the intra-day tick level
data for an NYSE traded fund. The Hurst
exponent was estimated to be 0.95, which
indicates a persistent time series. A Hurst
exponent value between 0.5 and 1.0 indi-
cates persistent behavior; the larger the
H value the stronger the trend.
CONCLuSION
The Hurst exponent is a useful statis-
tical method for inferring the properties
of a time series without making assump-
tions about stationarity. It is most useful
t i me s e R i e s
If you’re a operations researcher or analytics professional, looking for a new
position, INFORMS can help. Leading employers in operations research and
related fields will be at the INFORMS Job Fair during our Annual Meeting in
Phoenix, Arizona from October 14–17, 2012.
ARE YOU WAITING FOR THE PHONE TO RING?
Contact us today. You’ll be glad you did!
visit http://jps.informs.org
call 1-800-4INFORMs, or 443-757-3500
e-mail us at [email protected]
» BE PROACTIVE.
MEET POTENTIAL O.R. EMPLOYERS. Figure 2: An anti persistent time series (H = 0.043). Figure 3: A persistent time series (H = 0.95).
when used in conjunction with other
techniques, and has been applied in a
wide range of industries. For example
the Hurst exponent is paired with tech-
nical indicators to make decisions about
trading securities in financial markets;
and it is used extensively in the health-
care industry, where it is paired with
machine-learning techniques to monitor
EEG signals. The Hurst exponent can
even be applied in ecology, where it is
used to model populations. ❙
REF ERENCES
1. H.E. Hurst, 1951, “Long-term storage of
reservoirs: an experimental study,” Transactions of
the American Society of Civil Engineers, Vol. 116,
pp. 770-799.
2. Bo Qian, Khaled Rasheed, 2004, “Hurst
Exponent and fnancial market predictability,”
IASTED conference on “Financial Engineering and
Applications”(FEA 2004), pp. 203-209,
3. Mandelbrot, Benoit B., 2004, “The (Mis)Behavior
of Markets, A Fractal View of Risk, Ruin and
Reward,” Basic Books, 2004.
Subir Mansukhani is an innovation lead analyst with Mu
Sigma.
subscri be t o Anal yt i cs
It’s fast, it’s easy and it’s FREE! Just visit: http://analytics.informs.org/
www. i nf or ms . or g
The IBM Corporation has a long and deep
history as a company that both uses opera-
tions research and, more generally, advanced
analytics to run its business. It is also a place
where practitioners and innovators in the feld
have chosen to work for decades. Many im-
portant contributions to the feld have been
made over the years, and those contributions
continue to this day.
IBM is also one of the oldest, largest and
most historical companies in the country. The
company, founded 1911 as the Computing
Tabulation Recording (C-T-R) Corporation,
celebrated its 100th anniversary last year.
With more than 425,000 employees world-
wide, IBM is the second largest publically
traded technology company in the world (as
measured by market capitalization), the 31st
largest corporation in the Forbes ranking and
the 18th largest on the Fortune magazine list
of largest companies in the United States.
IBM has had many very pivotal and im-
portant leaders over the years, but among
the earliest and most colorful was Thomas J.
Watson Sr. He was the person who adopted
the motto “THINK,” and it is possible to trace
a century of innovation to the principles and
values created for the company at that time.
This focus on innovation can be seen by
the fact that the company holds more pat-
ents than any other U.S.-based technology
corporation and that many of its employees
have garnered important recognition for their
accomplishments (including five Nobel Prize
winners and too many other prizes to men-
tion in this short article). In addition, Thomas
J. Watson Sr. created the Watson Scientific
Computing Laboratory at Columbia Univer-
sity in 1945 that blossomed into the IBM Re-
search Division. Today, the IBM Research
Division has more than 3,000 researchers
globally in nine labs around the world, and
32 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
analytics at iBm
By aRnOlD
gReenlanD
One of the hallmarks
of the corporation has
been its ability to adapt
the business to remain
a vibrant and growing
entity.
technology and consulting giant boasts a long, rich history with
operations research and advanced analytics.
cOR p OR at e p R Of i l e
Headquartered in Armonk, N.Y., IBM holds more patents than any other U.S.-based technology company and has nine
research laboratories worldwide.
www. i nf or ms . or g 33 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
it is generally regarded as one of the
most critical components to the success
of the corporation over the decades.
TrANSFOrMATION OF ThE COMpANy
One of the hallmarks of the IBM Cor-
poration has been it ability to adapt the
business to remain a vibrant and growing
entity. While the company clearly began as
a manufacturer of business machines, it
has evolved into a company that provides
a combination of hardware, software and
services to its business and other organi-
zational clients. A major milestone in the
history of the company is typically referred
to as the “transformation,” which began in
the early 1990’s when Louis V. Gerstner
assumed leadership of the corporation.
Gerstner is credited with turning around
the fortunes of the company, which had
begun to suffer in the late 1980s and
early 1990s. He led the company into the
services and software businesses. What
is especially interesting, and gratifying to
the audience for this profle, is that opera-
tions research remains a pivotal compo-
nent of each and every one of the three
core IBM businesses today.
Operations research (O.R.) and ad-
vanced analytics can be found in critical
places throughout the company’s many
divisions, but the root of the power in O.R.
can clearly be seen within IBM Research
Division. The importance of operations
research within the IBM Corporation be-
gan early in the development of the IBM
Research organization as evidenced by
the leadership of the eminent operations
researcher Ralph E. Gomory as director
of the IBM Research Lab, then a senior
executive in the corporation. Over the
years, IBM Research has been the home
for outstanding operations researchers
such as Phil Wolfe, Alan Hoffman, An-
drew Conn and Bill Pulleyblank, just to
name a few. This leadership in the feld of
O.R. continues today under the direction
of Brenda Dietrich, an IBM Fellow and VP
for Business Analytics and Mathematical
Sciences. Brenda leads an organization
of more than 300 researchers, many of
them O.R. specialists. Brenda has also
been a leader in the O.R. feld as a re-
cent president of INFORMS and a regular
participant in INFORMS meetings, com-
mittees and leadership.
IBM researchers continue to make crit-
ical contributions in the feld of O.R. and
leadership for the corporation in its focus
on business analytics and optimization.
whErE’S ThE O.r.?
So where’s the operations research at
IBM? The answer is, simply, everywhere.
We already mentioned the pivotal place
of operations research and advanced
analytics within the IBM Research Di-
vision, but it is also prominent in every
division of the corporation: hardware,
software and services. Consider frst
the company’s hardware division. O.R.
has been a key part of the manufactur-
ing and fabrication capabilities of the
company for its entire history, but most
recently it can seen impacting the oper-
ations of the chip fabrication operations
that are focused in Fishkill, NY. Using a
wide range of O.R. tools, including opti-
mization, simulation and data analysis,
IBM has tightened operations, increased
cOR p OR at e p R Of i l e
Get “Suite” on
INFORMS Journals
Sign-up for unlimited access to all
12 INFORMS online journals and
save over $150.
It just makes sense.
• The cost of Pubs Suite OnLine is the same
as subscribing to 4 journals.
• Subscribe to Pubs Suite today for only $99
to access all online journal content back to
volume 1, issue 1.
• Upgrade today at INFORMS Member
Self-Service Center

www.informs.org/members-pubssuite
Pubs
hel p promote Anal yt i cs
It’s fast and it’s easy! Visit: http://analytics.informs.org/button.html
www. i nf or ms . or g 34 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
productivity and improved proft by the
activities of this operation.
As IBM has evolved into one of the
largest software companies in the world,
it has also included advanced analytics
– and operations research in particular –
in its growth and development plans. IBM
has historically made software contribu-
tions to O.R., including the early develop-
ment of optimization tools such as MPSX
and OSL. More recently the corporation
has chosen to grow its portfolio of analyt-
ics software tools by acquisition, spending
more than $14 billion in acquisitions in this
area alone. Key acquisitions in this area
include ILOG, which contains the popu-
lar and widely used CPLEX optimization
product. In addition the ILOG suite has vi-
sualization, constraint programming, sup-
ply chain planning tools and a robust and
popular rules engine (JRULES).
In the statistical and data mining area,
IBM recently purchased SPSS, a fully
functional statistical analysis package
along with a highly competitive data and
text mining capability in the SPSS Mod-
eler product (previously known as Clem-
entine). IBM also purchased the business
intelligence tool, COGNOS, one of the
leaders in that feld. In addition to these
major analytics tools, the corporation has
purchased scores of smaller and more
niche analytics and modeling tools that
are being integrated into the portfolio, as
well as having IBM Research participate
in the development of new software tools
internally.
Like the other divisions of the compa-
ny, the Services Division houses a small
capability in operations research and ad-
vanced analytics. While this capability
is relatively new, it is growing rapidly as
the corporation has created the Business
Analytics and Optimization (BAO) prac-
tice within the Global Business Services
(GBS) Division. While IBM has actually
been a services provider for decades,
these services were often bundled with
the hardware sales. With the acquisition
of PricewaterhouseCoopers Consult-
ing in 2002, the company made a strong
commitment to growing its services busi-
ness, and the addition of BAO as a core
service provided one of the critical recent
decisions. An article published in the IN-
FORMS digital journal, Analytics (analyt-
ics-magazine.org), entitled the “Analytics
Journey” [1], Lustig, et. al., describe in
detail the BAO practice. The key ideas
driving the development of this business
focus were articulated in the wildly pop-
ular and motivating book by Davenport
and Harris [2], which has turned out to be
a source of much of the language of the
analytics revolution that we are currently
all involved in.
The key message is that one of the core
services that companies want and need
are precisely the solutions that O.R. practi-
tioners have been developing and applying
for years to their businesses. This business
has grown in dramatically in the last couple
of years to the point where IBM includes
some 8,000 professionals among the ranks
of those that deliver BAO services. These
professionals include a broad range of an-
alytical talents from those who help client
development strategies, those who build
the IT infrastructures required to implement
the analytics capabilities, to, of course, the
advanced analytics practitioners who build
and apply the operations research and ad-
vanced analytics tools that are the heart of
the analytics revolution.
ACCOMpLIShMENTS
As a result of the legacy of opera-
tions research within the company (and
the promise of continued growth in the
future), IBM can point to a great many
accolades and accomplishments. Some
have already been mentioned above, as
they relate to IBM Research. Among its
many other noteworthy accomplishments
in the O.R. arena, IBM won the coveted
Edelman Prize from INFORMS in 1999
for work in 1999 entitled “Extended En-
terprise Supply Chain Management at
IBM Personal Systems Group and Other
Divisions.” IBM has also been on six other
Edelman fnalist teams. In addition, IBM
won the INFORMS Prize in 2000, and
numerous IBMers have won other IN-
FORMS prizes including the von Neuman
Prize and the Wagner Prize (won just last
year in the area of workforce planning).
In summary, IBM is a company that has
had a long and distinguished history and a
company that continues to adapt, grow and
contribute to the technology marketplace.
It is also a company that leveraged and re-
lied on operations research for its past suc-
cesses, and it is a company that has placed
a big bet on business analytics and optimi-
zation as a critical strategy for the future. ❙
Arnie Greenland ([email protected]) is a
distinguished engineer and technical executive at the
IBM Corporation. Greenland’s current role within the
corporation is serving as an analytics subject matter
expert for IBM’s Business Analytics and Optimization
practice, where he develops new or proposed business
solutions, regularly interacts with clients and mentors
emerging technical leaders. Early in his career he held
an academic appointment at George Mason University
in the Mathematical Sciences Department where he
taught courses in operations research, statistics and
mathematics. He holds a Ph.D. in mathematics. He is a
senior INFORMS member and a founding member of the
INFORMS Certifcation Committee.
cOR p OR at e p R Of i l e
REF ERENCES
1. Lustiz, I., Dietrich, B., Johnson, C., Dziekan, C.,
“The Analytics Journey,” analytics-magazine.org,
2010.
2. Davenport, T.H. and Harris, J.G., “Competing on
Analytics: The New Science of Winning,” Harvard
Business School Press, 2007.
www. i nf or ms . or g
How long should I wait for my bags after
a fight before deciding they are lost? This
practical problem is the focus of this issue’s
analysis.
First, a question so basic it is rarely asked:
Why wait for your bags at all? Why not, say,
go get your rental car, have a leisurely dinner,
make some phone calls and then come back
to collect your bags from the unclaimed pile
against the wall at your leisure? The answer
is, as far as I can tell, in two parts: First, people
like being united with their “stuff.” This is emo-
tional and not amenable to analysis. Second,
people are anxious to get their bags because
of the real or perceived risk of theft.
Without thinking too hard, we nominate two
“end post” strategies:
• Greedy-Service: Immediately report to the
baggage offce when you get off the plane.
File a missing bag report. If the person
who is taking the report notes (correctly)
that as all the bags have not come off the
carousel yet, you cannot know if your bag
is truly missing, you threaten to call their
manager. Upon fling your claim, if your bag
has arrived, you tear up the report and be
on your way.
• Greedy-Bags: Immediately report to the
baggage carousel when your fight lands.
Do not leave until you either have your
bags in hand or they turn off the lights for
the night and send everyone home.
Neither of these strategies are sophisticat-
ed. They are two extremes that help to scope
the problem. So let’s think about the risks that
are associated with each strategy. First, the
Greedy-Service strategy is best if your bag is
actually lost. If your bag is not lost, you lose
both time that you spend waiting to fle the
claim, as well as the possibility of your bag be-
ing stolen if it is not lost. If you choose Greedy-
Bags, you lose the value of your time waiting
for the bag, which may never arrive.
Let’s do some math and be a bit more rigor-
ous in our thinking. Suppose that you are con-
strained to choosing between these two very
simple strategies but with a twist: You get to
decide before landing that at some time, call it
, that you will switch from the baggage strat-
egy to the service strategy. In plain language,
you will abandon the baggage carousel for the
claims line.
First, we need to defne some parameters.
Let: be the probability that a bag is lost. The
bag will actually be lost before the plane takes
off from it’s origin, but the fact of loss will not
be revealed until it fails to arrive. Let be the
per hour risk of theft, expressed in bags stolen
per hour. Let be the value of your bag, and
its contents. Let be the value of your time,
expressed in dollars per hour. Finally, let
be the time that the carousel runs. We assume
for simplicity’s sake that if you move from the
baggage carousel to the line, that you will stay
at the claims line until .
The resulting expression is best understood
in pieces. Let’s talk about the things that can
happen:
1. My bags could be present and they could
get stolen when I walk away to join the
line to report them as missing. This has a
cost of , where the
middle term is the cumulative probability
that an exponentially distributed random
event occurs in .
2. My bags could be present and I will waste
time standing in line when I could have
been waiting for the bag. This has a cost of
.
3. My bags could be absent and I could
waste time waiting at the carousel that I
could have been fling a claim. This has a
cost of .
Now, all we have to do is put it all together.
Since these are expressed in terms of costs,
we want to:
With the restriction on meaning that
you cannot decide to get in the baggage line
35 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
Baggage claim
By haRRisOn schRamm
what strategy should an
arriving airline passenger
pursue if their bags are
missing?
f i ve - mi nu t e analys t
36 | a na ly t i cs - maga z i ne . or g
before the plane has landed, and if
you wait until the baggage carousel
stops, you will be forced to get in the
claims line.
Now it looks like an ugly mess;
we’re looking for the global minimum of
a non-linear function. It’s not that bad,
however, and a “quick and dirty” way to
solve this is to use “line search,” which
is a grown-up version of “too hot- too
cold.” You may also differentiate this
equation, which is what supports my
conclusions below.
For example, if my bag and contents
are worth $170, I bill at $100 per hour,
the maximum time the carousel runs is
20 minutes and the baggage theft risk
is one bag per hour, the optimum solu-
tion is to wait for 20 minutes, i.e., wait
for my bag, or in our original parlance
“Greedy-Bags.”
It turns out that in order to drive the
solution away from , one of the
following has to happen:
• The bag must be practically
worthless. Not really practically
applicable because people do not
bother to transport worthless bags.
Even if the market value of the bags
is zero, they usually have some
value to the passenger.
• There must be a high probability of
baggage loss. The historical rates of
baggage loss in the United States
are below 1 percent.
• You must have a very high billing
rate. People with these fee
schedules typically do not collect
their own bags!
So, we’ve done some math and dis-
covered something that we knew all
along: For most people, the best thing
to do when getting off the airplane is to
simply wait for their bags.
Bonus: This problem has many oth-
er dimensions; I have not done it jus-
tice in such a short note. For example,
I did not treat the information gained by
seeing others still waiting for their bags,
which would tend to make you be will-
ing to wait longer. ❙
Harrison Schramm ([email protected])
is a military instructor in the Operations Research
Department at the Naval Postgraduate School in
Monterey, Calif. The author acknowledges Ned
Dimitrov for posing this problem. He is an INFORMS
member.
f i ve - mi nu t e analys t
A membership in INFORMS provides...

• Online access to the latest in operations research
and advanced analytics techniques
• Subscriptions to INFORMS Publications
• Networking Opportunities available at INFORMS
Meetings and Communities
• Education programs around the world to
enhance your professional development
and growth!
Visit http://join.informs.org
Why go it alone?
and become part of the largest
professional society in the world for
professionals in the field of operations
research (O.R.), management science,
and business analytics.
Join INFORMS
Request a no-obligation infORms member Benefits packet
For more information, visit: http://www.informs.org/Membership
found in the “Instructions, rules and notes”
section at the end of this article and at www.
puzzlor.com.
puzzLE #1 - SurvIvOr
Getting lost while hiking in the wilderness
is a dangerous situation. And making your
way back to civilization is a difficult task that
uses up resources quickly. What you de-
cide to take with you while making the jour-
ney back to civilization can determine life or
death.
Table 1 shows all of the items that are
available to you that will aid you in your hike
out of the wilderness. Containers of Food
and Water will give you energy, Shelter will
www. i nf or ms . or g
If you’re a fan of the “Thinking Analytically”
column or just puzzles in general, a challeng-
ing new contest will put your analytics skills to
the ultimate test.
Below are a series of fve puzzles that fo-
cus on common problems encountered in the
mathematics, statistics and operations re-
search felds. Each puzzle requires a specifc
solution methodology that you have likely ac-
quired during your education and career.
Once you have solved each of the puzzles,
place the answers into the corresponding lati-
tude and longitude as indicated:
Input solution here:
Latitude _ _. _ _ _ _ _,
A B . C D E F G
Longitude -_ _. _ _ _ _ _
H I . J K L M N
A $100 cash prize is hidden at these co-
ordinates. There is no entry fee and anyone
can participate. Additional information can be
protect you from the elements, and Defense
will protect you from wild animals. Each item
has a weight indicated by the red number and
each item has survival points indicated by
the green number. You must take exactly one
item from each of the four categories (Food,
Water, Shelter, Defense). Unfortunately, the
backpack you have has a maximum capac-
ity of 25 kg. Your chance for survival is cal-
culated by adding all of the survival points
together from the items you choose to take
with you.
Question: What is the maximum number of
survival points you can achieve?
Answer No. 1: _ _ points
H A
puzzLE #2 - ChOOSE yOur CrEw
Successfully navigating the waters dur-
ing sea voyages is a challenging task. A
37 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
the 2012 analytics
treasure hunt
By JOhn tOczek
five analytics puzzles, one coordinate, one $100 cash prize.
Brought to you by www.puzzlor.com
t h i nk i ng analy t i cal ly
Table 1: Necessities for survival in the wilderness.
Table 1: Can you pick the right crew for the voyage?
www. i nf or ms . or g 38 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
captain’s most important decision is se-
lecting the right crew for the voyage. A
mix of different skill sets is required to
sail the ship efficiently, navigate to the
destination and fish for food along the
way.
Table 1 shows a list of crew mem-
bers who are available for you to hire
for the voyage. Each crew member de-
mands a salary for the voyage and has
different skill levels of Fishing, Sailing
and Navigation.
In order for your journey to be suc-
cessful, you must have a cumulative skill
of 15 or more in each of the three skill
categories from all of your chosen crew
members. You may choose as many
crew members as you like.
Question: What is the minimum achiev-
able cost for the voyage?
Answer No. 2: _ , _ _ _
I J B C
puzzLE #3 - TrAvELINg SpACEMAN
prOBLEM
Table 1 shows a list of galaxies that
you, as the traveling spaceman, wish to
visit. The table shows the coordinates in
three-dimensional space where each gal-
axy exists in the universe.
Question: In what order should you vis-
it each galaxy to minimize the traveled
distance? You must visit each galaxy
and you cannot visit any galaxy more
than once.
Answer No. 3 : _ -> _ -> _ -> _ ->
K E L
_ -> Z8 -> L3
D
puzzLE #4 - CONNECTED AND
INFECTED
Disease can spread quickly in a glo-
balized world where increased travel
raises the opportunity for transmission.
Figure 1 shows a modified map of the
world divided into five areas. The arrows
show the potential path for the spread
of disease. For example, the Blue area
can transmit infections directly to the
Red and Violet areas but not to the
Yellow or Green areas. Each area can
also transmit disease to its own popula-
tion, indicated by the arrows that circle
back onto themselves.
The populations for each area are
as follows: Blue has 0.5 billion people,
Red has 0.5 billion people, Violet has
4.0 billion people, Yellow has 1.0 billion
people and Green has 0.5 billion peo-
ple. Initially everyone is healthy except
for 10,000 people in the Blue area who
are infected with a virus.
Each month, 6.1 percent of the
infected people transmit the virus to
healthy people along each transmis-
sion route. For example, at the end
of month 1, the Red and Violet ar-
eas will each have 610 infected peo-
ple (10,000 infected people from the
Blue area x 6.1 percent) and the Blue
area will have 10,610 infected people
(10,000 who were originally infected
plus 10,000 x 6.1 percent from infect-
ing its own population). Infected peo-
ple stay infected and no deaths occur
in this scenario.
Question: After how many months is
at least half of the world’s population
infected?
Answer No. 4: _ _ Months
F M
t h i nk i ng analy t i cal ly
Table 1: So many galaxies to visit, so little time.
Figure 1: How long until half the world is infected?
subscri be t o Anal yt i cs
It’s fast, it’s easy and it’s FREE! Just visit: http://analytics.informs.org/
hel p promote Anal yt i cs
It’s fast and it’s easy! Visit: http://analytics.informs.org/button.html
39 | a na ly t i cs - maga z i ne . or g
puzzLE #5 - rELIEF MISSION
Coordinating relief efforts after ca-
tastrophes such as civil unrest and
natural disasters can be a logistically
complex challenge. Delivering relief to
people in need is the immediate focus
of any disaster management plan.
The map in Figure 1 shows the
locations of 20 villagers, each repre-
sented by a “hut” icon. The villagers
are in need of relief supplies con-
tained in the crates attached to para-
chutes. There are two identical relief
packages available. The only delivery
option is by air drop. Each package
can be dropped on any cell.
After the crates are dropped, each
villager will walk to the nearest drop
location to pick up relief supplies. Use
a direct line between cells to calcu-
late travel distance. For example, the
distance between a1 and a2 is 1km
and the distance between A1 to B2
is 1.41 km. Assume that each crate
contains an unlimited amount of relief
supplies.
Question: Which two drop locations
will minimize the total distance that all
villagers must travel?
Answer No. 5: _ _ , h _
G N
INSTruCTIONS, ruLES AND NOTES
Instructions
Solve the fve puzzles in this docu-
ment and place their answers in the lat-
itude and longitude spaces indicated.
When the puzzles are correctly com-
pleted, and the answers are placed in
t h i nk i ng analy t i cal ly
Figure 1: Which two drop locations are optimal?
Joi n t he Anal yt i cs Sect i on of I NFORMS
For more information, visit: http://www.informs.org/Community/Analytics/Membership
40 | a na ly t i cs - maga z i ne . or g
the corresponding latitude and longi-
tude slots provided at the beginning of
this article, the coordinate will indicate
the location of a hidden treasure.
Rules
1. You can reach the coordinates
easily by traveling on public roads
and land. The treasure is NOT
hidden on private property. If your
coordinates take you to private land
(residential homes, businesses,
commercial buildings, etc.) your
solution(s) are incorrect.
2. Minimal effort is required to retrieve
the treasure from its hiding place.
You do not need to dig or move
anything.
3. The frst person (or team) to retrieve
the treasure is the winner. Once
the treasure is found, the contest is
over.
4. The contest is brought to you by
www.puzzlor.com. Contestants
participate at their own risk. The
author and www.puzzlor.com,
Analytics magazine, INFORMS
and Lionheart Publishing are
not responsible for any injuries,
damages or costs incurred while
participating in this contest.
5. If you fnd the treasure, send an
e-mail to [email protected]
indicating that you have found
it so the puzzlor website can be
updated.
6. Have fun and be safe!
Notes
• A GPS enabled device will be
helpful in leading you to the
treasure’s coordinates.
• The contest is active until someone
correctly completes all of the
puzzles and retrieves the treasure
from its hiding space. Notifcation
will occur on the puzzlor website
(www.puzzlor.com) when someone
has retrieved the treasure.
• Additional questions can be sent to
[email protected]. ❙
John Toczek is the manager of Decision Support and
Analytics for ARAMARK Corporation in the Global Risk
Management group. He earned a bachelor’s degree in
chemical engineering at Drexel University (1996) and
a master’s degree in operations research from Virginia
Commonwealth University (2005). He is an INFORMS
member.
t h i nk i ng analy t i cal ly
Check for the posts about the meeting from our team of
bloggers at http://meetings.informs.org/Analytics2012. Blog
See what was happening by viewing tweets on http://twitter.com.
Search for hashtag “#analytics2012” to associate them with
the conference.
We’ve collected the tweets and posted them at
http://meetings.informs.org/Analytics2012.
It’s never been easier to revisit key topics from the
2012 INFORMS Analytics Conference
Applying Science to the Art of Business
USINESS ANALYTICS &
PERATIONS RESEARCH
INFORMS CONFERENCE ON
RECAP THE MEETING with ANALYTICS
CONFERENCE ONLINE TOOLS
Check out the meeting photos on flickr!
Connect with conference participants and discuss key
topics at the Practice Conference LinkedIn Group. Visit
http://meetings.informs.org/Analytics2012
to link to the page.
Request a no-obligation infORms member Benefits packet
For more information, visit: http://www.informs.org/Membership
www. i nf or ms . or g
The 2012 INFORMS Annual Meeting will
be held Oct. 14-17 at the Phoenix Convention
Center in the heart of downtown Phoenix. The
Convention Center was recently expanded
with the surrounding Sonoran Desert serving
as its architectural inspiration from colors to
textures to materials. Two major conference
hotels, the Hyatt Regency and Wyndham, are
just steps away.
Within walking distance attendees will fnd
numerous restaurants and attractions including
the home venues of the Arizona Diamondbacks
baseball team, Phoenix Suns basketball team,
Arizona Theatre Company (the Tony award win-
ning musical “Next to Normal” will be playing dur-
ing the conference), Phoenix Opera Company,
Phoenix Symphony and the Orpheum Theatre.
The Orpheum’s Spanish Baroque Re-
vival design dates back to vaudeville and
has earned it a place on the National Regis-
ter of Historic Places. The internationally re-
nowned Heard Museum, just a short ride from
downtown on the Light Rail will be presenting
“Beyond Geronimo: the Apache Experience”
for those interested in learning the truth about
the Apache culture and heritage.
The Organizing Committee has several ex-
citing events planned for the conference. The
general reception will draw on the theme of
Route 66 to explore the wonders of traveling
cross country and the splendor of Northern Ari-
zona that inspired numerous artists from Ansel
Adams’s photographs to Zane Grey’s and Tony
Hillerman’s novels to the lyrics of the Eagles
(remember that “corner in Winslow, Ariz.”?).
The Society for Medical Decision Making is
holding its annual meeting at the same location
from Oct. 17-20, and Oct. 17 will have several
overlapping sessions open to registrants from
both conferences. In keeping with that theme,
41 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
phoenix to host infORms
annual meeting
By ROn askin
speakers will cover
OR/ms contributions,
opportunities and
challenges in a variety
of areas including
intelligence and defense,
healthcare, homeland
security, manufacturing,
logistics and the future
of computing in an
increasingly networked
world.
cOnf e R e nce p R e v i e w
The Phoenix Convention Center will be the site of the 2012 INFORMS Annual Meeting.
42 | a na ly t i cs - maga z i ne . or g
a conference tour will be available to
visit the Mayo Clinic physical simula-
tion center.
The meeting will offer an exciting
blend of networking and educational
opportunities. As in the recent past, the
INFORMS membership meeting will be
held on Saturday (Oct. 13), a welcome
reception on Sunday (Oct. 14) and subdi-
vision meetings on Monday and Tuesday
(Oct. 15 and 16) followed by the general
reception on Tuesday evening (Oct. 16).
A stimulating list of plenary and key-
note speakers from academia, gov-
ernment and industry will share their
knowledge and perspectives. Planned
speakers will cover OR/MS contribu-
tions, opportunities and challenges in
a variety of areas including intelligence
and defense, healthcare, homeland se-
curity, manufacturing, logistics and the
future of computing in our increasingly
networked world. The Edelman award
reprise and Wagner Prize presenta-
tions will be included in the program.
For the most current information,
visit http://meetings2.informs.org/phoe-
nix2012/. Information on the Phoenix
area is available at www.visitphoenix.
com/index.aspx. ❙
Ron Askin ([email protected]) is the general chair
of the 2011 INFORMS Annual Meeting. He is a senior
INFORMS member.
cOnf e R e nce p R e v i e w
speakers will cover
OR/ms contributions,
opportunities and
challenges in a variety
of areas including
intelligence and defense,
healthcare, homeland
security, manufacturing,
logistics and the future
of computing in our
increasingly networked
world.
subscri be t o Anal yt i cs
It’s fast, it’s easy and it’s FREE!
Just visit: http://analytics.informs.org/
www. i nf or ms . or g 43 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
SCIENCE OF BETTEr: pODCASTS
Gain insights from experts on how math,
analytics, and operations research affect
organizations like yours in these 20-30
minute podcasts conducted by INFORMS
Director of Communications Barry List.
Visit www.scienceofbetter.org/podcast
Gary Cokins, SAS consultant
Mystery of Dying Industry Giants
Recorded May 24, 2012
Wally Hopp and Roman Kapuscinski
Does American Manufacturing Have a
Future?
Recoded May 11, 2012
U.S. Army Major Rob Dees
Measure of a Soldier
Recorded April 27, 2012
Theresa Kushner, the Senior Director
of Customer Intelligence at Cisco
Marketing Analytics at Cisco
Recorded March 30, 2012
Sheldon Jacobson, University of
Illinois Urbana-Champaign
March Madness O.R. Style
Recorded March 16, 2012
Renee Adams, University of New South
Wales and Patricia Funk, Universitat
Pompeau Fabra and Barcelona
Graduate School of Economics
Beyond the Glass Ceiling
Recorded March 2, 2012
Atanu Basu, Ayata
Analytics and the future of healthcare
Recorded February 3, 2012
Stefanos Zenios and Constantia
Petrou, Stanford University
Helping patients decide
Recorded January 19, 2012
Emanuel Derman, Columbia
University
Models Behaving Badly
Recorded January 5, 2012
Michael Johnson,
University of Massachusetts
Boston
Defender of Neighborhoods
Recorded December 27, 2011
Laura McLay, Virginia
Commonwealth
Punk Rock Blogging
Recorded December 8, 2011
Alfred Blumstein, Carnegie Mellon
Crime and Redemption
Recorded November 23, 2011
Gary Gaukler, Texas A&M
Still Existent Threat
Recorded October 28, 2011
Kevin Maney, Journalist
Two-second Advantage
Recorded October 6, 2011
Michael Rappa, North Carolina State
University
Teaching analytics
Recorded September 15, 2011
Ryan W. Buell, doctoral student
and Assoc. Prof. Michael I. Norton,
Harvard Business School
Do Customers Hate Waiting?
Recorded September 2, 2011
Doug Samuelson, InfoLogix
OR in the ER
Recorded August 19, 2011
Alberto Galasso, University of Toronto
& Timothy Simcoe , Boston University
Overconfdence: A Secret Asset for
CEOs?
Recorded August 8, 2011
Pinar Keskinocak & Julie Swann, GA
Institute of Technology
The Humane Face of Analytics
Recorded July 21, 2011
Alessandro Acquisti, Carnegie
Mellon University
Privacy on the Internet?
Recorded July 7, 2011
US Air Force Maj. Gen. (Ret.) Richard
O’Lear, Co-Chair, Defense Science
Board Task Force on Defense
Intelligence
Strengthening Defense Intelligence
Recorded June 24, 2011
l e aR ni ng R e s Ou R ce s
infORms’ library
of audio and video
presentations
BaRRy list
Request a no-obligation
infORms member Benefits packet
For more information, visit:
http://www.informs.org/Membership
www. i nf or ms . or g 44 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
INFOrMS vIDEO LEArNINg CENTEr
View these free, on-demand pre-
sentations, complete with slides, from
INFORMS renowned meetings and con-
ferences. Visit http://livewebcast.net/
INFORMS_Video_learning_Center.
ANALyTICS CONFErENCE 2012
2012 Edelman Award Presentations
Chris Goossens, TNT; Hein Fleuren,
Tilburg University; Davide Chiotti,
TNT; Marco Hendriks, TNT
Supply Chain-Wide Optimization at
TNT Express
Frederic Deschamps, Carlson; Pelin
Pekgun, JDA Software; Suresh
Acharya, JDA Software; Kathleen
Mallery, Carlson
Carlson Rezidor Hotel Group
Maximizes Revenue through Improved
Demand Management and Price
Optimization
Greg Burel, CDC; Eva K. Lee,
Georgia Institute of Technology
Centers for Disease Control and
Prevention: Advancing Public Health
and Medical Preparedness with
Operations Research
Sofa Archontaki, Danaos; Takis
Varelas, Danaos; Iraklis Lazakis,
Danaos; Evangelos Chatzis, Danaos
Operations Research in Ship
Management: Maximizing Fleet-Wide
Revenue Routing at Danaos
Suresh Subramanian, HP; Prasanna
Dhore, HP; Girish Srinivasan, HP;
David Hill, HP
Hewlett-Packard: Transformation
of HP’s Business Model through
Advanced Analytics and Operations
Research
Karl Kempf, Intel; Feryal Erhun,
Stanford University; Robert Bruck,
Intel
Optimizing Capital Investment
Decisions at Intel Corporation
Panel Discussion:
Diego Klabjan, Northwestern
University; Thomas Olavson,
Google; Blake Johnson, Stanford
University; Daniel Graham, Teradata;
Michael Zeller, CEO, Zementis, Inc.
Innovation and Big Data: Panel
Discussion
ANNuAL MEETINg 2011
Wagner Prize Presentations
Karl Kempf, Intel; Evan Rash, Intel
Product Line Design and Scheduling at
Intel
Adeline Kuo, Analytics Operations
Engineering, Inc.; Anjuli Kannan,
Analytics Operations Engineering,
Inc.; Gerald van den Berg, Princeton
University
iSchedule to Personalize Learning
(Analytics Operations Engineering, Inc.
on behalf of New York City Department
of Education)
Vijay Mookerjee, University of Texas
at Dallas; Subodah Kumar, Texas
A&M University; Radha Mookerjee,
University of Texas at Dallas
To Show or Not To Show: Using
User Profling to Manage Internet
Advertisement Campaigns at Chitika
Stefan Spinler, WHU - Otto
Beisheim School of Management;
Paul Kleindorfer, INSEAD; Andrei
Neboian, WHU--Otto Beisheim
School of Management; Alain Roset,
Groupe La Poste
Fleet Renewal with Electric Vehicles at
La Poste
Eva Lee, Georgia Institute of
Technology; Chien-Hung Chen,
Georgia Institute of Technology
Designing Guest Flow and Operations
Logistics for the Dolphin Tales (Georgia
Tech on behalf of The Georgia
Aquarium)
ANALyTICS CONFErENCE 2011
2011 Edelman Award Presentations
John Bear, Midwest ISO; Richard
Doying, Midwest ISO; Mingguo
Hong, Midwest ISO
Midwest ISO Unlocks Billions in
Savings through the Application of
O.R. to Energy and Ancillary Services
Markets
l e aR ni ng R e s Ou R ce s
Additional Videos
More great learning videos based
on award-winning presentations at
INFORMS conferences are available
in the Video Learning Center.
www. i nf or ms . or g
You’ve probably heard a lot about dig data,
largely as a result of the fact that the technology
– in the shape of ultra-fast processors and data
interfacing systems – has come of age, meaning
that companies can harness the power that big
data brings to the better technology table.
However, plenty of confusion persists about
what dig data is and how it helps the average
hard-pressed company professional.
At its most basic, big data is an umbrella term
for any pro-active use of available data for the pur-
poses of improving services and customer satis-
faction. In this context, a major focus has been
placed on data warehousing and data mining for
better analytics on a company’s customers, as
well as their product or service consumption.
The underlying premise is that the data re-
quired for analysis is available to the company
concerned and is in a format that is easily ac-
cessible. The data should also, of course, be
reliable enough to support analytics.
But wait – as the TV advert says – there’s
more, as big data generates information
that analysts call BI (business intelligence),
which, unlike the raw materials used in
manufacturing processes, can be used, re-
used and re-used again.
BI is now a must-have feature of mod-
ern management. A 2011 IBM survey found
that 83 percent of chief information offcers
view BI as their top priority for enhancing
competitiveness.
Until just a few years ago, businesses
tended to limit and even block the data they
supplied to people outside of their day-to-day
environment and bring information inside. But
the arrival of big data – and the raw BI it gener-
ates – allows companies to do the reverse and
share their inside data with customers and see
what they do with it.
This is, in essence, how the more effcient
businesses communicate with their customers
on social networking site and services such as
Facebook and LinkedIn.
As a result, many organizations are fnding
that a high percentage of BI now resides out-
side the structured environment, meaning that
businesses have to change the methodology
by which they get data, which can pose sig-
nifcant technical challenges. Assuming these
technical challenges can be overcome and the
underlying big data supporting the organisa-
tion’s information resource is reliable, then we
can start to crunch the available information.
For most applications, historical data meets
the reliability criterion, but technical limitations re-
main, caused by the fact that a lot of data is being
exchanged across networks at lightening speeds
and with service lifetimes that are often reduced
to the time it takes to download an app.
And here is where it gets interesting, as our
observations suggest that the optimum level of
customer satisfaction occurs at the “moments
of truth” where the customer interacts with the
service, a concept made famous by Jan Carl-
son of Scandinavian Airlines.
When it comes to communication networks,
these moments of truth are occurring in real
time and at very high speed. Put simply, this
means that, whilst a great deal of effort can
be expended on analyzing and understanding
a customer’s service consumption history, the
real measure of customer satisfaction is how
well the service provider can satisfy customer
needs at the moment of truth.
Let’s think about what this means for the
underlying IT system. While the concept of
big data is relatively easy to understand, the
very term itself is likely to send shivers down
the spine of the IT professional, for the simple
reason that moving large volumes of data in
real time means that one or more technology
bottlenecks will be encountered.
These bottlenecks differ between organiza-
tions, but the central focus is that there needs
to be real-time data analysis of customer ser-
vice usage available to management in order
that they can assemble the key performance
indicators (KPI) that modern business plan-
ning now thrives on.
Questions that need to be answered include:
Did the customer get the service they wanted
and was it provided satisfactorily? Were there
any delays or resends? Were there any issues
with congestion that prevented the customer
45 | a na ly t i cs - maga z i ne . or g a na ly t i cs | J U ly / aU gU s t 2012
understanding the
challenges and
opportunities of big data
By Dan-JOe BaRRy
many organizations
are finding that a high
percentage of Bi now
resides outside the
structured environment,
meaning that businesses
have to change the
methodology by which
they get data, which can
pose significant technical
challenges.
v i e wp Oi nt
46 | a na ly t i cs - maga z i ne . or g
getting the service when they needed it
as fast as they needed it?
The only way to collect and analyze
this information is to complete the pro-
cess in real time as the moment of truth
unfolds.
The bottom line here is that captur-
ing this information on customer ser-
vice usage and network performance
is the crucial front-end to understand-
ing if the service delivery is living up
to expectations. It’s important to un-
derstand that this information is not
only useful for understanding the cur-
rent situation, but can also be used
to enhance the historical information
that KPI projections are often based
on.
By historical information, we mean
data on which services customers are
using, as well as when and for how long,
so allowing pro-active service providers
to change their service offering to better
suit customers’ behaviour. From a tech-
nology perspective, this is the back-end
we traditionally understand as support-
ing big data, but the essential front-end
is real-time data collection on those
crucial “moments of truth,” which, in the
end, determine customer satisfaction. ❙
Dan-Joe Barry is vice president of marketing with
Napatech (www.napatech.com).
v i e wp Oi nt
OPTIMIZATION
www.gams.com
Europe
GAMS Software GmbH
Eupener Strasse 135-137
50933 Cologne, Germany
phone
+49-221-949-9170
fax
+49-221-949-9171
mail
[email protected]
web
http://www.gams.com
USA
GAMS Development
Corporation
1217 Potomac Street, NW
Washington, DC 20007, USA
phone
+1-202-342-0180
fax
+1-202-342-0181
mail
[email protected]
web
http://www.gams.com
GAMS Integrated Developer Environment for editing,
debugging, solving models, and viewing data.
High-Level Modeling
The General Algebraic Modeling System (GAMS)
is a high-level modeling system for mathemati-
cal programming problems. GAMS is tailored for
complex, large-scale modeling applications, and
allows you to build large maintainable models that
can be adapted quickly to new situations. Models
are fully portable from one computer platform to
another.
State-of-the-Art Solvers
GAMS incorporates all major commercial and
academic state-of-the-art solution technologies for
a broad range of problem types.
The Network Enabled Optimization
System
The Network Enabled Optimization System (NEOS)
started at the Argonne National Laboratory in the
1990s. Since 2010, it has been hosted at the University of Wisconsin in the Wisconsin Institutes for
Discovery. The NEOS server (www.neos-server.org) is on the cutting-edge of optimization software, and
allows optimization problems to be solved automatically with minimal input from the user.
• The site hosts both academic and commercial solvers. Problems can be submitted from modeling sys-
tems such as GAMS and AMPL, and also described using a number of other input formats.
• Drawing from computational resources from CHTC (chtc.wisc.edu), NEOS has completed more than
100,000 jobs in the frst four months of 2012 alone.
• The system also has a NEOS Guide (www.neos-guide.org) containing information about solvers and
optimization software, and a growing collection of optimization case studies.
Users are encouraged to help the NEOS team to enhance this educational outreach activity.
For more information please visit: http://www.neos-server.org
Jobs Submitted by Solver
Type (Q1)
Input Type (Q1, 2012)
Other
3,929
AMPL
33,000
GAMS
40,000
S
B
B
G
u
r
o
b
i
S
N
O
P
T
B
o
n
m
in
M
I
N
T
O
M
I
N
O
S
K
N
I
T
R
O
B
A
R
O
N
P
A
T
H
X
p
r
e
s
s
.
.
.
O
t
h
e
r
30,000
20,000
10,000
0
Job Submission Options
Browser
Email Client
Custom App
GAMS
AMPL
Web Interface
Email
XML-RPC
User
capturing information on
customer service usage
and network performance
is the crucial front-end
to understanding if the
service delivery is living
up to expectations.
hel p promote Anal yt i cs
It’s fast and it’s easy! Visit:
http://analytics.informs.org/button.html

Sponsor Documents

Or use your account on DocShare.tips

Hide

Forgot your password?

Or register your new account on DocShare.tips

Hide

Lost your password? Please enter your email address. You will receive a link to create a new password.

Back to log-in

Close