-
8/10/2019 1-s2.0-S0007681314000901-main.pdf
1/11
Supply chain analytics
Gilvan C. Souza
Kelley School of Business, Indiana University, Bloomington, IN
47405, U.S.A.
1. Why analytics in supply chainmanagement?
The
supply
chain
for
a
product is
the
network offirms and facilities involved in the
transformationprocess from raw materials to a product and inthe
distribution of that product to customers. Ina supply chain, there
are physical, financial, and
informational flows among different firms. Supplychain analytics
focuses on the use of informationandanalytical
tools
to
make
better
decisions regard-ing material flows in the supply chain.
Putdifferently, supply chain analytics focuses on
analytical
approaches
to
make
decisions
that
bettermatch supply and demand.Well-planned and implemented
decisions con-
tribute directly to the bottom line by loweringsourcing,
transportation, storage, stockout, anddisposal costs. As a result,
analytics has historicallyplayed a significant role in supply chain
manage-ment, starting with military operations during and
after World War IIparticularly
with the
develop-ment of the simplex method for solving linear
pro-gramming by George Dantzig in the 1940s. Supplychain analytics
became more ingrained in decisionmaking with the advent of
enterprise resource plan-ning (ERP) systems in the 1990s and more
recentlywith big data applications, particularly in descrip-tive
and
predictive
analytics,
as
I describe
withsome examples in this article.
Business
Horizons
(2014)
57, 595605
Available online at www.sciencedirect.com
ScienceDirectwww.elsevier.com/locate/bushor
KEYWORDS
Supply chainmanagement;Analytics;Optimization;Forecasting
Abstract In this article, I describe the application of advanced
analytics techniquesto supply chain management. The applications
are categorized in terms of descrip-tive,
predictive,
and
prescriptive
analytics
and
along
the
supply
chain
operationsreference (SCOR) model domains plan, source, make,
deliver, and return. Descriptiveanalytics applications center on
the use of data from global positioning systems(GPSs), radio
frequency identification (RFID) chips, and data-visualization tools
toprovide
managers
with
real-time
information
regarding
location
and
quantities
ofgoods in the supply chain. Predictive analytics centers on
demand forecasting atstrategic, tactical, and operational levels,
all of which drive the planning process insupply chains in terms of
network design, capacity planning, production planning,
andinventory
management.
Finally,
prescriptive
analytics
focuses
on
the
use
of
mathe-matical optimization and simulation techniques to provide
decision-support toolsbuilt upon descriptive and predictive
analytics models.
#
2014
Kelley
School
of
Business,
Indiana
University.
Published
by
Elsevier
Inc.
Allrights reserved.
E-mail
address:
[email protected]
0007-6813/$
see
front
matter # 2014 Kelley School of Business, Indiana University.
Published by Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.bushor.2014.06.004
http://dx.doi.org/10.1016/j.bushor.2014.06.004http://dx.doi.org/10.1016/j.bushor.2014.06.004http://dx.doi.org/10.1016/j.bushor.2014.06.004http://dx.doi.org/10.1016/j.bushor.2014.06.004http://dx.doi.org/10.1016/j.bushor.2014.06.004http://dx.doi.org/10.1016/j.bushor.2014.06.004http://dx.doi.org/10.1016/j.bushor.2014.06.004http://dx.doi.org/10.1016/j.bushor.2014.06.004http://www.sciencedirect.com/science/journal/00076813mailto:[email protected]://dx.doi.org/10.1016/j.bushor.2014.06.004http://dx.doi.org/10.1016/j.bushor.2014.06.004mailto:[email protected]://www.sciencedirect.com/science/journal/00076813http://dx.doi.org/10.1016/j.bushor.2014.06.004http://crossmark.crossref.org/dialog/?doi=10.1016/j.bushor.2014.06.004&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.bushor.2014.06.004&domain=pdf
-
8/10/2019 1-s2.0-S0007681314000901-main.pdf
2/11
The Supply Chain Operations Reference (SCOR)model developed by
the Supply Chain Council(www.supply-chain.org) provides a good
frameworkfor classifying the analytics applications in supplychain
management.
The
SCOR
model
outlines
fourdomains of supply chain activities: source, make,deliver,
and return. A fifth domain of the SCOR
modelplanis behind
all
four
activity
domains.Furthermore, a key input of the supply chain plan-ning
process is demand forecasting at all timeframes: long, mid, and
short term with planninghorizons of years, months, and days,
respectively.Table 1 illustrates different decisions in each of
thefour SCOR domains that can be aided by analytics.These decisions
are further classified into strategic,tactical, and operational
according to their timeframe.Analytics techniques can be
categorized into
three types: descriptive, predictive, and prescrip-tive.
Descriptive
analytics
derives
information
from
significant amounts of data and answers the ques-tion of what is
happening. Real-time informationabout the location and quantities
of goods in thesupply chain
provides
managers
with
tools
to
makeadjustments to delivery schedules, place
replenish-mentorders, place emergency orders, change
trans-portation
modes,
and
so
forth.
Traditional
datasources include global positioning system (GPS) dataon the
location of trucks and ships that containinventories, radio
frequency identification (RFID)data originating
from
passive
tags
embedded
in
pallets (even at the product level), and transactionsinvolving
barcodes. Information is derived from thevast amounts of data
collected from these sourcesthrough data visualization, often with
the help ofgeospatial mapping
systems.
RFID
is
a
significantimprovement over barcodes because it does notrequire
direct line of sight. Accurate inventory re-
cords are
critical
in
supply
chains
as
they
triggerregular replenishment orders and emergency orderswhen
inventory levels are too low. Although RFIDtechnology helps in
significantly reducing the fre-quency of manual inventory reviews,
such reviewsare still needed because of data inaccuracy due to,for
example, inventory deterioration or damage oreven tag-reading
errors.Predictive analytics in supply chains derives de-
mand forecasts
from
past
data
and
answers
thequestion of what will be happening.Prescriptive analytics
derives decision recom-
mendations
based
on
descriptive
and
predictive
analytics models and mathematical optimizationmodels. It answers
the question of what should behappening. Arguably, the bulk of
academic re-search, software,
and
practitioner
activity
in
supplychain analytics focuses on prescriptive analytics.In Table
2, I provide a summary of analytics
techniquesdescriptive,
predictive,
and
prescrip-tiveused in supply chains in terms of the four
SCORdomains of source, make, deliver, and return. Ielaborate on
Table 1 and Table 2 in the nextsections.
Table
1.
SCOR
model
and
examples
of
decisions
at
the
three
levels
SCOR
Domain
Source
Make
Deliver
Return
Activities Order and receivematerials andproducts
Schedule andmanufacture, repair,remanufacture,
orrecycle
materialsand products
Receive, schedule,pick, pack, andship
orders
Request, approve, anddetermine disposal ofproducts
and
assets
Strategic
(time frame:years)
Strategicsourcing
Supply chainmapping
Location of plantsProduct line mixat plants
Location ofdistribution centers
Fleet planning
Location of returncenters
Tactical(time frame:months)
Tactical sourcingSupply chaincontracts
Product linerationalization
Sales andoperations planning
Transportation anddistribution planning
Inventory policiesat locations
Reverse distributionplan
Operational
(time
frame:days)
Materialsrequirementplanningand inventoryreplenishmentorders
Workforce schedulingManufacturing,
ordertracking, and scheduling
Vehicle routing(for
deliveries)
Vehicle routing (forreturns
collection)
Plan Demand forecasting (long term, mid term, and short
term)
596 G.C. Souza
http://www.supply-chain.org/http://www.supply-chain.org/
-
8/10/2019 1-s2.0-S0007681314000901-main.pdf
3/11
2. Plan: Demand forecasting usingpredictive analytics
Demand forecasting is a critical input to supplychain
planning.
Different
time
frames
for
demand
forecasting require different analytics techniques.Long-term
demand forecasting is used at the stra-tegic level and may use
macro-economic data, de-mographic trends, technological trends,
andcompetitive intelligence. For example, demand fac-tors for
commercial aircraft at Boeing include ener-gy prices, discretionary
spending, populationgrowth, and inflation, whereas demand factors
formilitary aircraft include geo-political changes, con-gressional
spending, budgetary constraints, andgovernment regulations (Safavi,
2005). Causal fore-casting methodscalled
such
because
they
analyze
the underlying
factors that
drive
demand
for aproductare used at this level. Analytics
causalforecasting
methods
include
linear, non-linear,and logistic regression.To illustrate demand
forecasting for tactical and
operational supply chain decisions, consider theproduction
planning
process
for
an
original
equip-ment manufacturer (OEM) such as Whirlpool. At theproduct
family level (e.g., refrigerators), the salesand operations
planning
(S&OP)
process
uses
aggre-gate demand forecasts in monthly time buckets toestablish
aggregate production rates, aggregatelevels of inventories, and
workforce levels. Theaggregate plan is revised on a rolling basis
as newdata is available. The S&OP plan, as well as morerefined
demand forecasts at the stock-keeping unit(SKU) level, is used to
derive the master productionschedule (MPS), which details weekly
productionquantities at the SKU level for a typical planninghorizon
of 812 weeks. The MPS and the bill ofmaterials are
then
used
to
plan
production
andsourcing at the part level through a materials re-quirement
planning (MRP) system that is embeddedinmost ERP
software.Time-based demands for parts
are derived from time-based demands for theSKUs that use those
parts, so parts have dependentdemand. In contrast, SKUs have
independentdemand. Demand forecasts for items subject to
inde-pendent
demand require
predictive
analytics techni-
ques,whereasforecasts fordependent demanditemsare obtained
directly from the MRP system. Demandforecasts for independent
demanditemsarealsousedtoplan for inventory safety stocks at other
locations,such as distribution centers and retailers.Demand
forecasting for independent demand
items isusuallyperformedusingtime-seriesmethods,forwhich the
only predictor of demand is time. Time-series methods include
moving average, exponentialsmoothing, andautoregressive models. For
example,Wintersexponential smoothingmethodincorporatesboth
trend
and
seasonality
and can
be
used
for
both
short-term and
mid-term forecasting.
In
an
autore-gressive model, demand forecast in one period is
aweighted sum of
realized demands
in
the previousperiods.Mid-term forecasting can also benefit
from
causal forecasting methods, especially in non-manufacturing
industries
or
the
manufacturing
ofnon-discrete items. For example, in order to fore-cast monthly
demand for truckload (TL) freightservices, Fite,
Taylor,
Usher,
English,
and
Roberts(2002) considered 107 economic indexes as poten-tial
predictors, including the purchasing managersindex, the Dow Jones
stock index, the consumergoods production index, automotive dealer
sales,U.S. exports, the producer commodities price indexfor
construction materials and equipment, interestrates, and gasoline
production. They used stepwiseregression to identify the most
relevant indexes andfound parsimonious models for predicting TL
de-mand for specific industries and regions. Their mod-el only
predicts
industry-wide
demand
for
TLservices (nationally or by region); the connectionto demand
forecasts at the firm level was madeusing historical market
shares.
Table
2.
Analytic
techniques
used
in
supply
chain
management
Analytics
Techniques
Source Make Deliver Return
Descriptive Supply
chain
mapping Supply
chain
visualization
Predictive Time series methods (e.g., moving average,
exponential smoothing, autoregressive models) Linear, non-linear,
and logistic regression Data-mining techniques (e.g., cluster
analysis, market basket analysis)
Prescriptive Analytic hierarchy process Game theory (e.g.,
auction design,contract design)
Mixed-integer linearprogramming (MILP)
Non-linear programming
Network flowalgorithms
MILP Stochasticdynamicprogramming
Supply chain analytics 597
-
8/10/2019 1-s2.0-S0007681314000901-main.pdf
4/11
Data mining has also been used for demand fore-casting in
conjunction with traditional forecastingtechniques (Rey, Kordon,
& Wells, 2012). Usually,the data-mining step precedes the use
of causalforecasting techniques
by
finding
appropriate
de-mand drivers (i.e., independent variables) for aproduct that
can be used in regression analysis.
For example,
Dow
Chemical
uses
a
combination
ofdata mining and regression techniques to forecastdemand at the
strategic and tactical levels (e.g.,identifying demand trends),
which is useful for itspricingstrategy and for configuring and
designing itssupply chain to respond to these trends
(Rey&Wells,2013). Data-mining methods usually involve
cluster-ing techniques. So, if a retailer finds out, for exam-ple,
that demand for cereal is strongly related tomilk sales,
then
the
retailer
may
build
a
causalforecasting model that predicts cereal sales withmilk
sales as one of the predicting variables. Marketbasket analysis
is
a
specific
data-mining
technique
that provides an analysis of purchasing patterns atthe
individual transaction level, so a retailer cananalyze the
frequency with which two product cat-egories
(e.g.,
DVDs
and
baby
products)
are
pur-chased together. Lift for a combination of items isequal to
the actual number of times the combina-tion occurs
in
a
given
number
of
transactions
dividedby the predicted number of times the combinationoccurs if
items in the combination were indepen-dent. Lift values above 1
indicate that items tend tobe purchased
together.
This
kind
of
analysis
can
beuseful when building causal regression models for
demand
forecasting. It
can
also
aid
in
promotionactivitiesbecause the retailer can predict how
muchsales of Product 1 would increase if there is apromotion for
Product 2 if the two products areoften purchased together.
3. Source
3.1. Source: Strategic decisions
Strategic sourcing is the process of evaluating andselecting key
suppliers. There is limited use ofanalytics for
strategic
sourcing
in
practice
eventhough academics prescribe the use of
sophisticatedmulti-criteria decision-making techniques such
asanalytic hierarchic process (AHP). AHP decomposesa complex
problem (e.g., selecting a supplier amonga diverse set) into more
easily comprehended sub-problems that can be analyzed separately.
In thesupplier-selection problem, these sub-problemsmight include
distinct evaluations of factors likecost, quality, delivery speed,
delivery reliability,volume flexibility, product mix flexibility,
and sus-tainability. These evaluations are then weighed.
Firms are very familiar with their first-tier sup-pliers (i.e.,
those that directly supply them) andperhaps their second-tier
suppliers (i.e., those thatsupply first-tier suppliers), but some
of their lower-tier suppliers
may
be
unknown.
A
recent
example
isthe November 2012 fire at the Bangladesh factorythat killed
more than 100 workers. An audit of the
factory by
Walmart
in
2011
ruled
it
out
as
a
supplier.However, one of Walmarts suppliers continued
tosubcontract work to that factory (Tsikoudakis,2013). The threat
of disruptions like natural disas-ters, social and political
unrest, and major strikesmakes it imperative for firms to map their
supplychains. For example, Cisco (2013) uses supply chainmapping
and enterprise social networking to iden-tify its vulnerabilities
to supply chain disruptions aswell as
to
collaborate
with
its
suppliers
and
part-ners. The open source tool sourcemap.com, devel-oped at the
Massachusetts Institute of Technology,allows one
to
visualize
and
map
a
supply
chain;
the
tool can also be used for purposes such as carbonfootprint
estimation. An example is shown inFigure 1.
3.2. Source: Tactical decisions
In
contrast
to
strategic sourcing,
tactical
sourcingrefers to the process of achieving
specificobjectivessuch as determining costs for parts,materials, or
servicesthrough structured procure-ment mechanisms
like
auctions.
The
central
prob-lem in procurement auctions centers on mechanism
design:
How
should
one
structure
the
rules
of
anauction so that bidders (i.e., suppliers) behave in amanner
that results in minimal procurement cost(and desired performance)
for the buyer? Auctionscan be open (i.e., bidders can view and
respond tobids) or
sealed
and
one
shot
or
dynamic
(whichoccurover several rounds of bidding). Government auc-tions
tend to be one-shot, sealed auctions, whereasopen, dynamic auctions
are common in industrialprocurement (Beil, 2010). Buyers must
consider thetotal procurement cost as bidders usually bid
oncontract payment terms only (e.g., unit cost). Ad-ditional
logistics
costs,
if
paid
by
the
buyer,
must
betaken into account in the bid price. The prescriptiveanalytics
used here is centered on game theory,which is used to determine
auction rules. Procure-ment auctions are widely used in practice.A
commonly used payment contract in sourcing is
wholesale price, via which the buyer (i.e., retailer)pays the
seller (i.e., manufacturer) a fixed price perunit. Under this
contract, retailers are exposed todemand risk: they bear the entire
costs of over-stocking and therefore have an incentive to stockless
than what is optimal for the supply chain as a
598 G.C. Souza
-
8/10/2019 1-s2.0-S0007681314000901-main.pdf
5/11
whole. Recognizing this, academics have used acombination of
game theory and statistics to pre-scribe more sophisticated
contracts that will im-prove product availability in retailers. For
example,in a buy-back contract, retailers can return unsoldunits to
the manufacturer and receive a partialrefund. Although
such
contracts
can
improve
supplychain performance, the wholesale price contract isstill
widely used, perhaps due to its simplicity.
4. Make
4.1. Make: Strategic decisions
Network design determines the optimal location andcapacity of
plants, distribution centers (DCs), andretailers. The simplest form
of the network design
problem can be illustrated when deciding where tobuild DCs that
serve as intermediary stockingand shipping points between existing
plants andretailers. This problem is formulated as a mixed-integer
linear program (MILP). Data requirementsinclude yearly aggregate
demands for the productfamily at
each
retailer,
plant
capacities,
unit
ship-ping costs between each pair of locations, and theannual
fixed cost of operating a DC at each potentiallocation. Decision
variables include the quantity toship between locations and binary
variables thatindicate if each DC should be open or closed.
Theobjective function minimizes total shipping andfixed DC costs.
Constraints ensure that demand ismet at all locations, that
companies only ship prod-ucts from a DC if it is open, and that all
plantcapacities are respected. The solution providesthe location
(i.e., where to open the DCs) as well
Figure 1. Example of supply chain mapping using
sourcemap.com
Source:
free.sourcemap.com/view/6585/
Supply chain analytics 599
-
8/10/2019 1-s2.0-S0007681314000901-main.pdf
6/11
as the allocation of plants to the DCs, the allocationof DCs to
retailers, and the capacity of each DC.Variations of this simple
MILP formulation include
multipleproducts, transportation capacitiesbetweenlocations,
multiple
transportation modes
betweenlocations, a multi-year planning horizon,
multipleechelons (i.e., tiers in the supply chain), demand
uncertainty,
supply
uncertainty,
and reverse
flows(e.g., the collection of used products for recyclingand
remanufacturing). When the problem incorpo-ratesmultipleproducts,
theanalysis alsoprovides theproduct mix at each plant. When many of
the varia-tions above are incorporated andthe problem is
large(e.g., thousands of retailers and potential DC andplant
locations), the problem may become too diffi-cult to solve to
optimality using off-the-shelf optimi-zation software.
Therefore,
many researchers
haveproposed well-performing heuristics, such as
geneticalgorithms, that ensure goodand sometimesoptimalsolutions.
Genetic
algorithms use
a
divide
and conquer (the feasible region) approach to findinga good
solution to the MILP as opposed to optimalbranch and bound
algorithms, which are combinato-rial in nature.Some of the data
necessary to perform such
analysis requires a preliminary level of analysis soit can
be
extracted,
cleaned,
and
aggregated
fromERP systems. Network design, however, is only per-formed
infrequently for each firm, including duringmergers and
acquisitions. As a result, it is not part ofstandard ERP
software.
Specialized
software
makesit easy to input this data, specify the constraints,
perform
the
optimization,
and
visualize
the
results,especially for large problems.
4.2. Make: Tactical decisions
We have
previously
described
the
S&OP
process,which is used for planning aggregate workforceand
inventory levels on a medium planning horizonbased on demand
forecasts, underlying costs, andactual sales. Academics have
proposed MILP modelsfor this process. For each month in the
planninghorizon, decision variables include the amount toproduce
for
each
product
family
using
regular
time,overtime, and subcontracting and the number ofworkers to be
hired and laid off. The objectivefunction minimizes total cost,
which comprises totalproduction cost (i.e., regular time, overtime,
andsubcontracting), total inventory cost, total wages,total hiring
cost, and total layoff cost. Constraintsmay come from, among
others, inventory and work-force balancing, regular production
capacity,and overtime production. Many practitioners userules-based
heuristics. For example, one heuristicis a level production
strategy, via which the firm
meets fluctuating demand byproducing at a constantrateandholding
inventory tomeet thepeakdemand.Alternatively, the firm can use a
chase strategy,adjusting workforce levels monthly
tomeetfluctuat-ing demand.
Firms
frequently
use
a
hybrid
strategybetween chase and
level.Productproliferationandmasscustomizationhave
been widely
documented
(e.g.,
Rungtusanatham
&Salvador, 2008). For product proliferation and
masscustomization, the plant must adapt from a massproduction
environmentdesigned for economiesof scale, with fewer products
produced in dedicatedlines and setup costs spread over long
productionrunsto a flexible production environment. Thisadaptation
is made possible with the aid of flexiblemanufacturing technology
or changes in the productand process
design that
support
a
postponementstrategy (Lee, 1996). In a postponement strategy,the
step in the manufacturing process in which prod-uct
differentiation
occursfrom
gray boxes
to
SKUsis located closer to thecustomer, whichallowsthe firm to
carry inventory of gray boxes instead ofSKUs, and thus lessens
differentiation time. Post-ponement mitigates the
negative
impacts
of
in-creased product proliferation, such as increasedforecasting
uncertainty at the SKU level; increasedinventory costs;
and
complexity costs,
such
as
re-search and development, testing, tooling,
returns,andobsolescence.Postponement requires changes inproduct and
process design, and itmay not be feasi-ble for
products
like
automobiles,
for
which
strictquality guidelines in final assembly preclude signifi-
cant
customization
at
dealers.
As
an
alternative,firmsmay increase supply chainperformance
throughproduct rationalization using analytics, as shown inTable
3.
4.3. Make: Operational decisions
Manufacturing scheduling is the last step in theplanning process
after MRP plans are released. AnMRP plan specifies quantities and
due dates for allparts. Scheduling then sequences the jobs
(i.e.,parts) by the different resources necessary
formanufacturing
the
part
in
order
to
meet
the
duedates. In general, there are n jobs to be scheduled inm
different resources, and the processing time, duedate, and weight
(i.e., priority) of each job in eachresource are known. This
problem takes differentforms depending on the decision makers
objective,the number of resources, and how the jobs areprocessed
with the resources. An objective functionminimizes the maximum
completion time, or themaximum lateness, across all jobs. There can
beprecedence relationships, setup times, or evensequence-dependent
setup times (i.e., when the
600 G.C. Souza
-
8/10/2019 1-s2.0-S0007681314000901-main.pdf
7/11
setup time for a job at a resource depends on theprevious job
there, such as in processing industrieslike the chemical industry).
Scheduling problemscan be formulated as MILPs, and the
combinatorial
nature of these problems makes them very hard tosolve to
optimality
for
large
problems.
As
a
result,significant effort has been devoted to finding
goodsolutions through heuristics because other compli-cations arise
in practice, such as adding new jobs totheexisting pool of
processing jobs as well as chang-ing priorities
and
preferences.
In
terms
of
software,some ERP systems have scheduling modules (e.g.,the
Applied Planning and Optimization module inSAP) that use genetic
algorithms to provide goodsolutions to MILPs found in determinist
scheduling.Thesealgorithms can provide good solutions to fairly
large problems,
such
as
1
million
jobs
over
1,000resources
(Pinedo,
2008).
There
are
a
few
compa-nies such as Taylor (www.taylor.com) that specializein
providing scheduling softwarewith many function-alities not
present in
ERP systems.
Although
thediscussion above has centered on manufacturingscheduling,
some of the same algorithms can be usedin other scheduling problems
like assigning gates atan airport
or
trucks
at
a
cross-docking
location.Workforce scheduling can be challenging for ser-
vice industries, such as call centers, hospitals, andairlines,
in which there is seasonal demand, not onlyfor time of the year
(common inmanufacturing), butalso for day of the week and hour of
the day. Acommon way of modeling these problems isby defining
tours.
A
tour
is
a
combination
of
timeblockswithin a day and within days of the week thatadd up to
the necessary work hours per employee.An example of a tour would be
Monday, 8 a.m.1p.m.; Tuesday, 1 p.m.6 p.m.; Thursday, 8 a.m.6p.m.;
and Friday, 8 a.m.6 p.m. Tours should befeasible; for example, it
is not very convenient formost people
to
work
from
8
a.m.10
a.m.
andthen from 3 p.m.5:00 p.m. on the same day.
The decision maker needs demand forecasts foreach time block
(e.g., 12 p.m.1 p.m. on Monday),which can be obtained through
predictive forecast-ing models. This problem can be formulated as
an
MILP in which decision variables include the numberof
employees
assigned
to
each
tour
and
the
numberof employees necessary to meet demands withineach time
block. The objective function minimizestotal labor costs.
Complications, such as workerspreferences, multiple locations, task
assignments,and so
forth,
increase
the
size
of
the
MILP
model
tosuch an extent that heuristics are almost certainlyneeded.
Some ERP vendors have workforce sched-ulingmodules for specific
applications like retail andhospitality. There are also vendors for
industry-spe-cific software, such as call centers and health
care
providers.
Many
airlines,
which
are
heavy
analyticsusers, have
developed
their
own
scheduling
algo-rithms.
5. Deliver and return
5.1. Deliver and return: Strategicdecisions
In Section 4, I presented the network design problemof planning
the location of DCs and return centers.Another strategic decision
here is fleet planning,which can be described as the dynamic
acquisitionand divestiture of delivery vehicles to meet thedemand
for
deliveries
or
returns
collection.
Thisproblem is formulated as an MILP, or dynamic pro-gramming,
as in Table 4.
5.2. Deliver and return: Tactical decisions
In transportation and distribution planning, the
firmdistributes
a
set
of
products
from
source
nodes
(i.e.,supply points such as factories) to sink nodes
Table
3.
Product
rationalization
at
Hewlett-Packard
Hewlett-Packard (HP) has developed optimization tools for
product rationalization (Ward et al., 2010). One toolrequires
proposed new product line extensions to meet minimum complexity
return-on-investment (ROI)thresholds.
Complexity
ROI
is
defined
as
the
incremental
margin
minus
variable
complexity
costs,
divided
by
fixedcomplexity costs. Variable complexity costs are largely
driven by forecasting uncertainty and resulting increasedinventory
costs, whereas fixed complexity costs are driven by criteria such
as research and development, tooling,and manufacturing setup costs.
With another tool, HP uses a maximum flow algorithm on an existing
product line to
perform
product
rationalization.
The
tool
acknowledges
that
in
firms
with
configurable
product
lines,
someproducts, such as power supplies, may generate little
revenue on their own but are critical components for high-revenue
orders and for overall order fulfillment. Order coverage is defined
as the percentage of a given set of pastorders that can bemet from
the rationalized product portfolio. Similarly, revenue coverage is
the smallest portfolioof products that covers a given percentage of
historical order revenue. This optimization tool revealed how HP
canoffer only 20% of previously offered features in laptops and
reach 80% revenue coverage. After implementing therecommendations,
HP realized significantly reduced inventory costs and increased
gross margins.
Supply chain analytics 601
http://www.taylor.com/http://www.taylor.com/
-
8/10/2019 1-s2.0-S0007681314000901-main.pdf
8/11
(i.e., demand points such as retail locations)through
intermediary storage nodes (e.g., DCs).Thisproblem is solved using
a multi-commodity networkflowmodel,
which
is
a
linear
programming
formula-tion with a special structure. In the network
formu-lation, there can be multiple arcs between each pair
of nodes. Each arc represents a shippingmodewith agiven
capacity,
such
as
rail,
truckload
(TL),
less
thantruckload (LTL), and air. The amount to ship in eacharc in
the network for each commodity and timeperiod is
considered.
Constraints
include
capacity
ateach arc, time period, and node, as well as flow-balancing at
each node. Data requirements includeshipping costs in each arc,
forecasts of supply avail-able at each source node (provided by the
S&OPplan), point forecasts for demand at each sink node(from
predictive analytics models), and arc capaci-ties. Economies
of
scale
in
shipping
can
also
be
incorporated.
Problems
of
realistic
size have
thou-sands of nodes, resulting in millions of decisionvariables.
However, such problems can be solvedefficiently with numerical
algorithms based on thenetwork simplex
method,
which
is
embedded
insupply chain optimization software. Despite exten-sive
planning, disruptions (e.g., traffic, weather)and demand
uncertainty often require plan modifi-cation, and descriptive
analytics tools can be quitevaluable. For example, the Control
Tower descrip-tive analytics system allows Procter & Gamble
(P&G)to see
all
the
transportation
occurring
in
its nearsupply chain (i.e., inbound, outbound, raw materi-als,
and finished product). With this technology,P&G has reduced
deadhead movement (i.e., whentrucks travel empty or not optimally
loaded) by 15%and thus has reduced costs (McDonald, 2011).Another
important decision is determining
supply levels
at
nodes in
a
distribution
networkthat is, setting inventory policies. The science
forsetting inventory policies (i.e., reorder point andorder-up-to
level or order quantity) for a productat a single location, such as
a DC, is mature, evenwhen demand is uncertain and non-stationary
and
replenishment lead times are variable. Data re-quirements
include historic demand and forecastingdata, replenishment lead
times, the desired servicelevel (i.e.,
a
desired
fill
rate
or
stock-out
probabili-ty), holding cost, and the fixed cost of placing
areplenishment order. The inventory policy parame-
tersreorder point and order quantitycan becomputed
using
exact
algorithms
or
approximateformulas, which are embedded in most supply
chainsoftware, including in some ERP systems modules.More
often, the supply chain
has
multiple stockingpoints for the same product. For example, a
productcanbe stockedataDCandmultipledifferent retailersin different
regions. Although one can set inventorypolicies at each location
that use only local demandand replenishment lead-time information,
this localoptimization approach is not optimal for the supplychain.
Due
to risk
pooling,
it
may
be
optimal
to
have
some level of
inventory
at
the DC
so that
higher-than-normaldemand
inoneretailercanbebalancedagainstlower-than-normal demand at
another retailer. Thissituation calls for an integrated inventory
policy forthe entire
supply chain; the
theory
that prescribestheseinventorypolicies is calledmulti-echelon
inven-tory theory.Thecomplication inmulti-echelon
inven-torytheoryariseswhentheDCdoesnothavesufficientinventory
tomeetall incomingorders fromretailersata given period. In that
case, the optimal inventory-rationing policy is complex, and even
more so if thereare more
than two
echelons.
There
are,
however,several well-performing heuristics that are
computa-tionally simple, such as the guaranteed service
levelheuristic (Graves & Willems, 2000), which has
beenimplemented in software like Optiant. An example ofsuccessful
application is provided in Table 5.
5.3. Deliver and return: Operationaldecisions
The vehicle routing problem (VRP) optimizes thesequence of nodes
to be visited in a route, forexample, for a parcel delivery truck,
for a returns
Table
4.
Fleet
planning
for
Coca-Cola
Enterprises
Coca-Cola Enterprises (CCE) has started replacing some of its
fleet of diesel delivery trucks with diesel-electrichybrid vehicle
(HEV) trucks. How the company chooses to invest those dollars
depends on volatile fuel costs, usage-based
deterioration,
and
seasonal
demand.
Wang,
Ferguson,
Hu,
and
Souza
(2013)
have
provided
a
prescriptiveanalytics model that takes into consideration CCEs
historical maintenance costs, purchasing costs for both dieseland
HEV trucks, CCE demand data, and historical diesel price data to
calibrate a stochastic model that simulatesdiesel prices
dynamically. Using dynamic programming, the optimal policy is
obtained, at each period of a planning
horizon
and
for
each
realization
of
diesel
prices,
that
determines
how
many
trucks
of
each
type
(diesel
and
HEV)CCE should acquire and/or divest. Wang et al. found that at
the current outlook of diesel prices, CCE should includeboth HEV
(54%) and diesel trucks (46%) in its capacity portfolio. In this
regard, CCE could use HEV trucks to meet itsaverage baseline demand
and then deploy diesel trucks to supplement the delivery fleet
during peak demandseasons.
602 G.C. Souza
-
8/10/2019 1-s2.0-S0007681314000901-main.pdf
9/11
collection truck, or for both. The optimal sequencetakes into
account the distances between each pairof nodes; expected traffic
volume; left turns; andother constraints placed on the routes, such
asdelivery and pickup time windows. Known as thetravelling salesman
problem (TSP), the classical VRPproblem only
takes
into
account
the
distances
be-
tween each
pair
of
nodes:
In
what
sequence
shouldnodes be visited, ending at the same starting point?This
problem can be formulated as an MILP. The TSPproblem is
combinatorial in nature, and is hardto solve
beyond
a
few
thousand
nodes
(Funke
&Gruenert, 2005). Among others, complications suchas
multiple vehicles, vehicle capacities, tour-lengthrestrictions, and
delivery and pickup time windowsresult in an MILP that is very
difficult to solve, thusrequiring heuristic approaches. In addition
to heu-ristic approaches, vehicle-routing software incorpo-rates
descriptive
analytics,
as
shown
in
Table
6.
6. Modulating demand to matchcapacity: Revenue management
The SCOR model implicitly assumes that managersplan their
operationssource, make, deliver, andreturnbased on demand
forecasts. Therefore, theSCOR model plans capacity to match a given
de-mand. Industries
with
perishable
capacities,
likeairlines, hospitality, and transportation, must takea
reverse approach, so firms modulate their demand
to match their fixed capacity through prices andother mechanisms
that will be described next. Thisis known as revenue
management.Revenue management started in the airline in-
dustry after deregulation, with the problem of allo-cating seats
in a flight to fare classes. Allocationpolicies are
nested.
For
instance,
suppose
there
are
two fare
classes:
$150
(Fare
Class
1)
and
$90
(FareClass 2). The decision maker sets a booking limit forFare
Class 2 and then determines the booking limitof Fare Class 1 based
on the capacity of the flight.Data requirements
for
the
computation
of
bookinglimits include demand forecasts for the differentclasses
(as a probability distribution) at differenttimes before departure,
cancellation probabilities,up-selling probabilities (i.e., the
probability that acustomer will buy a higher fare if the lower fare
isunavailable), and fare values. The problem is sig-nificantly
more
complex
in
a
network.
For
example,one passenger goes from Indianapolis (IND) toNew York
(JFK), whereas another passenger goesfrom IND to Rochester (ROC)
via JFK. In this case,heuristic approaches, such as bid-price
controls, areused. The bid price for a resource (e.g., a seat in
aspecific flight IND-JFK) is the marginal cost to thenetwork of
consuming
one
unit
of
that
resource.When a customer demand arises (e.g., IND-ROC viaJFK),
then the demands revenue is comparedagainst the sum of bid prices
for all resources asso-ciated with the demand request (i.e., bid
prices fora seat IND-JFK and for a seat JFK-ROC). The demand
Table
5.
Multi-echelon
inventory
management
at
P&G
Before 2000, P&G used only single-location inventory models,
which optimize inventory levels locally given thatlocations own
replenishment lead time. However, starting in 20052006, P&G
started implementing multi-echeloninventory
models
based
on
the
guaranteed
service
level
heuristic
in
its
more
complex
supply
networks.
At
aparticular stage in the supply chain, inventory is set to meet
a desired service level based on a guaranteed deliverytime to the
customer (S), its own replenishment lead time when ordering from a
preceding stage (SI), and itsprocessing time (T). Essentially, the
method sets safety stock levels as if it was a single location with
a
replenishment
lead
time
of
SI
+
T
-
S.
Note
that
SI
for
a
stage
is
equal
to
S
for
a
preceding
stage.
Through
dynamicprogramming, the method finds the optimal S for each
stage to minimize holding costs across the supply chain.
Themulti-echelon supply chain approach to inventory management was
implemented at 30% of P&Gs locations usingOptiant software and
consequently saved the company $1.5 billion in inventory costs in
2009 compared to thesingle-location models previously in place
(Farasyn et al., 2011).
Table 6. Vehicle routing at Waste Management, Inc.
Waste Management, Inc. (WM) is a leading provider of solid waste
collection and disposal services. It has a fleet ofmore
than
26,000
vehicles
running
nearly
20,000
routes.
In
2003,
the
company
implemented
the
WasteRoutevehicle-routing software, which included GIS
capabilities and navigational capabilities, and integrated it with
arelational database containing customer information. An
origin-destination matrix was then developed thatconsidered
constraints such as time and distance traveled between any two
points, speed limits, and one-waystreets. By implementing the
combined prescriptive and descriptive analytics software, the firm
saved $44 millionin 2004.
Source:
www.informs.org
Supply chain analytics 603
http://www.informs.org/http://www.informs.org/
-
8/10/2019 1-s2.0-S0007681314000901-main.pdf
10/11
is accepted if the revenue is higher than the sum ofbid prices.
Bid prices can be approximated throughlinear
programming.
In capacity allocation, fare prices are given asthey are
determined by market forces. Another wayto manage uncertain demand
for fixed capacitybeit flight
seats,
hotel
rooms,
rental
cars,
or
inventoryin a retail environmentis through pricing. As ar-gued
by Talluri and Van Ryzin (2004, p. 175), thedistinction
between
quantity
and
price
controls
isnot always sharp (for instance, closing the availabil-ity of a
discount class can be considered equivalentto raising the products
price to that of the nexthighest class).
However, using
price
as
a
directmechanism to match demand with capacity is an
important
enough
practical
problem
to
merit
specialtreatment. Dynamic pricing has gained significanttraction
lately, particularly in retailing (i.e., mark-down pricing),
e-commerce, and even manufactur-ing (e.g., Fords offering of
incentives at its autodealers).The
key
is
to
find
a
good
predictive
demandmodel: At price p, what is the expected demandd(p) for the
product? Demand models may be linear(d(p) = a-bp), exponential
(i.e., constant elastici-ty), logit (i.e., S-curve), or
discrete-choice. Thereare many vendors of dynamic pricing software,
andsoftware calibrates the demand models using histor-ical
point-of-sale
data.
In
addition,
dataon
availableinventories is necessary for the
price-optimizationalgorithm. Different price-optimization
algorithmsareembedded in these packages based on non-linearand
dynamic programming.
7. Conclusion
Supply chain management is a fertile area for theapplication of
analytics techniques, which has his-torically been the case through
the use of operations
research, particularly linear programming andoptimization. For
example, inventory theory is morethan 50 years old, and there were
significantcontributions to production planning in the
1980s.Therefore, analytics
in
supply
chain
management
isnot new. More recent applications include the inte-gration of
price analytics and supply chain manage-
ment in
the
field
of
revenue
management,
for
whichthe problem revolves around managing demand inan
environment with fixed and perishable capacity.Revenue management
research and practice (par-ticularly inmanufacturing) is relatively
new becausemany demand models can only be calibrated
withsignificant amounts of data, which just recentlybecame
available from modern ERP systems andweb technologies.With
big
data,
new
opportunities
arise.
I haveheard consultants praising the potential of harness-ing
social networks for supply chain management,for example,
by
detecting
local
trends in
demand
to
adjust inventories and prices. There is indeed po-tential there,
although many firms still struggle tomatch basic supply and demand
in a world withincreased product
proliferation,
competition,
andglobalization (i.e., longer lead times). Among otherbenefits,
big data has the potential to improvedemand forecasting
methods,
detect
supply
chaindisruptions, and improve communications in supplychains
that are often global (see Table 7).
References
Ahuja,
R.,
Magnanti,
T.,
&
Orlin,
J.
(1993).
Network
flows:
Theory, algorithms, and applications. Upper Saddle River,NJ:
Prentice
Hall.Beil,
D.
(2010).
Procurement
auctions.
In
J.
J.
Hasenbein
(Ed.),2010
tutorials
in
operations
research:
Risk
and
optimization
in an uncertain world. Linthicum, MD: INFORMS.Campbell,
G.
(2009).
Overview
of
workforce
scheduling
software.Production and Inventory Management Journal, 45(2),
722.
Cisco.
(2013). Building a collaboration architecture for a global
supply chain. Retrieved June 14, 2013, from
http://www.cisco.com/en/US/solutions/collateral/ns340/ns858/c22-613680_sOview.html
Farasyn,
I.,
Humair,
S.,
Kahn,
J.,
Neale,
J.,
Rosen,
O.,
Ruark,
J.,
et
al.
(2011).
Inventory
optimization
at
Procter
&
Gamble:Achieving
real
benefits
through
user
adoption
of
inventorytools.
Interfaces, 41(1), 6678.Fite,
J.,
Taylor,
G.,
Usher,
J.,
English,
J.,
&
Roberts,
J.
(2002).Forecasting
freight
demand
using
economic
indices. Interna-
tional Journal of Physical Distribution and Logistics
Manage-
ment, 32(4), 299308.Funaki,
K.
(2009). State of the art survey of commercial software
for
supply
chain
design.
Retrieved
June
18,
2013,
from
http://www.scl.gatech.edu/research/supply-chain/GTSCL_scdesign_software_survey.pdf
Funke,
B.,
&
Gruenert,
T. (2005).
Local
search
for
vehicle
routingand
scheduling
problems:
Review
and
conceptual
integration.
Journal of Heuristics, 11(4), 267306.
Table
7.
Additional
information
on
analytics
techniques for supply chain management
General overview: Snyder and Shen (2011) Network
design:
Funaki
(2009) Auctions: Krishna (2002) Sales and operations planning:
Jacobs, Berry,Whybark, and Vollmann (2011)
Transportation
and
distribution
planning:
Ahuja,Magnanti, and Orlin (1993) Inventory management: Zipkin
(2000)Dynamic pricing and revenue management: Talluriand Van Ryzin
(2004)
Manufacturing scheduling: Pinedo (2008) Workforce scheduling:
Campbell (2009)
604 G.C. Souza
http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://www.cisco.com/en/US/solutions/collateral/ns340/ns858/c22-613680_sOview.htmlhttp://www.cisco.com/en/US/solutions/collateral/ns340/ns858/c22-613680_sOview.htmlhttp://www.cisco.com/en/US/solutions/collateral/ns340/ns858/c22-613680_sOview.htmlhttp://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://www.scl.gatech.edu/research/supply-chain/GTSCL_scdesign_software_survey.pdfhttp://www.scl.gatech.edu/research/supply-chain/GTSCL_scdesign_software_survey.pdfhttp://www.scl.gatech.edu/research/supply-chain/GTSCL_scdesign_software_survey.pdfhttp://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0040http://www.scl.gatech.edu/research/supply-chain/GTSCL_scdesign_software_survey.pdfhttp://www.scl.gatech.edu/research/supply-chain/GTSCL_scdesign_software_survey.pdfhttp://www.scl.gatech.edu/research/supply-chain/GTSCL_scdesign_software_survey.pdfhttp://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0030http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0025http://www.cisco.com/en/US/solutions/collateral/ns340/ns858/c22-613680_sOview.htmlhttp://www.cisco.com/en/US/solutions/collateral/ns340/ns858/c22-613680_sOview.htmlhttp://www.cisco.com/en/US/solutions/collateral/ns340/ns858/c22-613680_sOview.htmlhttp://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0015http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0010http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0005
-
8/10/2019 1-s2.0-S0007681314000901-main.pdf
11/11
Graves,
S.,
&
Willems,
S.
P. (2000).
Optimizing
strategic
safetystock
placement
in
supply
chains. Manufacturing and Service
Operations Management, 2(1), 6883.Jacobs,
R.,
Berry,
W.,
Whybark,
D.
C.,
&
Vollmann,
T. (2011).Manufacturing planning and control for supply chain
manage-
ment.
New
York:
McGraw-Hill.Krishna,
V.
(2002). Auction theory. New York: Academic Press.
Lee,
H.
(1996).
Effective
inventory
and
service
managementthrough
product
and
process
redesign. Operations Research,
44(1), 151196.McDonald,
R.
(2011,
November).
Inside
P&Gs
digital
revolution.McKinsey
Quarterly.
Available
at
http://www.mckinsey.com/insights/consumer_and_retail/inside_p_and_ampgs_digital_revolution
Pinedo,
M.
(2008). Scheduling theory, algorithms and systems
(3rd ed.).
New
York:
Springer.Rey,
T., Kordon,
A.,
&
Wells,
C.
(2012). Applied data mining for
forecasting using SAS. Cary, NC: SAS Institute.Rey,
T.,
&
Wells,
C.
(2013,
March
7).
Integrating
data
mining
andforecasting.
Analytics.
Retrieved
June
17,
2013,
from
http://www.analytics-magazine.org/march-april-2013/765-integrating-data-mining-and-forecasting
Rungtusanatham,
M.
J.,
&
Salvador,
F.
(2008).
From
mass
produc-tion
to
mass
customization:
Hindrance
factors,
structural
inertia,
and
transition
hazard. Production and Operations
Management, 17(3), 385396.Safavi,
A.
(2005).
Forecasting
demand
in
the
aerospace
anddefense
industry. Journal of Business Forecasting, 24(1),
27.Snyder,
L.
V.,
&
Shen,
Z-J.
M.
(2011).
Fundamentals
of
supply
chain theory. Hoboken, NJ: John Wiley & Sons
Inc.Talluri,
K.,
&
Van
Ryzin,
G.
(2004). The theory and practice of
revenue management. New York: Springer.
Tsikoudakis, M. (2013, January 13). Supply chain disasters
anddisruptions
can
cause
lasting
reputation
damage. Business
Insurance.
Retrieved June
14,
2013,
from
http://www.businessinsurance.com/article/20130113/NEWS06/301139968
Wang,
W.,
Ferguson,
M.,
Hu,
S.,
&
Souza,
G.
C.
(2013).
Dynamiccapacity
investment
with
two
competing
technologies.Manufacturing and service operations management,
15(4),616629.
Ward,
J.,
Zhang,
B.,
Jain,
S.,
Fry,
C.,
Olavson,
H.,
Amaral,
J.,et
al.
(2010).
HP
transforms
product
portfolio
managementwith
operations
research. Interfaces, 40(1), 1732.
Zipkin,
P. (2000). Foundations of inventory management.
New
York:
McGraw-Hill.
Supply chain analytics 605
http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0055http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0055http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0055http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0055http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0055http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0055http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0055http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0055http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0055http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0055http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://www.mckinsey.com/insights/consumer_and_retail/inside_p_and_ampgs_digital_revolutionhttp://www.mckinsey.com/insights/consumer_and_retail/inside_p_and_ampgs_digital_revolutionhttp://www.mckinsey.com/insights/consumer_and_retail/inside_p_and_ampgs_digital_revolutionhttp://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://www.analytics-magazine.org/march-april-2013/765-integrating-data-mining-and-forecastinghttp://www.analytics-magazine.org/march-april-2013/765-integrating-data-mining-and-forecastinghttp://www.analytics-magazine.org/march-april-2013/765-integrating-data-mining-and-forecastinghttp://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://www.businessinsurance.com/article/20130113/NEWS06/301139968http://www.businessinsurance.com/article/20130113/NEWS06/301139968http://www.businessinsurance.com/article/20130113/NEWS06/301139968http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0120http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0120http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0120http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0120http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0120http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0120http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0120http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0120http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0120http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0120http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0120http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0120http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0120http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0120http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0120http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0115http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0110http://www.businessinsurance.com/article/20130113/NEWS06/301139968http://www.businessinsurance.com/article/20130113/NEWS06/301139968http://www.businessinsurance.com/article/20130113/NEWS06/301139968http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0100http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0095http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0090http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0085http://www.analytics-magazine.org/march-april-2013/765-integrating-data-mining-and-forecastinghttp://www.analytics-magazine.org/march-april-2013/765-integrating-data-mining-and-forecastinghttp://www.analytics-magazine.org/march-april-2013/765-integrating-data-mining-and-forecastinghttp://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0075http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0070http://www.mckinsey.com/insights/consumer_and_retail/inside_p_and_ampgs_digital_revolutionhttp://www.mckinsey.com/insights/consumer_and_retail/inside_p_and_ampgs_digital_revolutionhttp://www.mckinsey.com/insights/consumer_and_retail/inside_p_and_ampgs_digital_revolutionhttp://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0060http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0055http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0050http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045http://refhub.elsevier.com/S0007-6813(14)00090-1/sbref0045
