PDF A handbook on multi-attribute decision-making methods omid bozorg-haddad download

Page 1


Visit to download the full and correct content document: https://ebookmass.com/product/a-handbook-on-multi-attribute-decision-making-metho ds-omid-bozorg-haddad/

More products digital (pdf, epub, mobi) instant download maybe you interests ...

Biofuels for a More Sustainable Future: Life Cycle Sustainability Assessment and Multi-Criteria Decision Making 1st Edition Jingzheng Ren (Editor)

https://ebookmass.com/product/biofuels-for-a-more-sustainablefuture-life-cycle-sustainability-assessment-and-multi-criteriadecision-making-1st-edition-jingzheng-ren-editor/

Marketing Management: A Strategic Decision-Making Approach 8th Edition

https://ebookmass.com/product/marketing-management-a-strategicdecision-making-approach-8th-edition/

Analytics Enabled Decision Making Vinod Sharma

https://ebookmass.com/product/analytics-enabled-decision-makingvinod-sharma/

Principles of Classroom Management: A Professional Decision-Making Model (7th Edition)

https://ebookmass.com/product/principles-of-classroom-managementa-professional-decision-making-model-7th-edition/

Connected Planning: A Playbook for Agile Decision Making 2nd Edition Ron Dimon

https://ebookmass.com/product/connected-planning-a-playbook-foragile-decision-making-2nd-edition-ron-dimon/

Statistics for Business: Decision Making and Analysis 3rd Edition

https://ebookmass.com/product/statistics-for-business-decisionmaking-and-analysis-3rd-edition/

Scientific Models and Decision Making 1st Edition Eric Winsberg

https://ebookmass.com/product/scientific-models-and-decisionmaking-1st-edition-eric-winsberg/

Big Data: An Art of Decision Making Eglantine Schmitt

https://ebookmass.com/product/big-data-an-art-of-decision-makingeglantine-schmitt/

Decision Making in Veterinary Practice Barry Kipperman

https://ebookmass.com/product/decision-making-in-veterinarypractice-barry-kipperman/

AHandbookonMulti-AttributeDecision-MakingMethods

WileySeriesinOperationsResearchandManagementScience

OperationsResearchandManagementScience(ORMS)isabroad, interdisciplinarybranchofappliedmathematicsconcernedwithimproving thequalityofdecisionsandprocessesandisamajorcomponentoftheglobal modernmovementtowardtheuseofadvancedanalyticsinindustryand scientificresearch.The WileySeriesinOperationsResearchandManagement Science featuresabroadcollectionofbooksthatmeetthevariedneedsof researchers,practitioners,policymakers,andstudentswhouseorneedto improvetheiruseofanalytics.Reflectingthewiderangeofcurrentresearch withintheORMScommunity,theSeriesencompassesapplication, methodology,andtheoryandprovidescoverageofbothclassicalandcutting edgeORMSconceptsanddevelopments.Writtenbyrecognizedinternational expertsinthefield,thiscollectionisappropriateforstudentsaswellas professionalsfromprivateandpublicsectorsincludingindustry, government,andnonprofitorganizationwhoareinterestedinORMSata technicallevel.TheSeriesiscomprisedoffoursections:Analytics;Decision andRiskAnalysis;OptimizationModels;andStochasticModels.

AdvisoryEditors • DecisionandRiskAnalysis

GilbertoMontibeller, LondonSchoolofEconomics

GregoryS.Parnell, UnitedStatesMilitaryAcademyatWestPoint

FoundingSeriesEditor

JamesJ.Cochran, UniversityofAlabama

Analytics

YangandLee • HealthcareAnalytics:FromDatatoKnowledgetoHealthcare Improvement

ForthcomingTitles

Attoh-Okine • BigDataandDifferentialPrivacy:AnalysisStrategiesfor RailwayTrackEngineering

KongandZhang • DecisionAnalyticsandOptimizationinDiseasePrevention andTreatment

DecisionandRiskAnalysis

Barron • GameTheory:AnIntroduction, SecondEdition Bozorg-Haddad,Zolghadr-Asli,andLoáiciga • AHandbookon Multi-AttributeDecision-MakingMethods

Brailsford,Churilov,andDangerfield • Discrete-EventSimulationandSystem DynamicsforManagementDecisionMaking

Johnson,Keisler,Solak,Turcotte,Bayram,andDrew • DecisionSciencefor HousingandCommunityDevelopment:LocalizedandEvidence-Based ResponsestoDistressedHousingandBlightedCommunities

MislickandNussbaum • CostEstimation:MethodsandTools

ForthcomingTitles

AlemanandCarter • HealthcareEngineering

OptimizationModels

Ghiani,Laporte,andMusmanno • IntroductiontoLogisticsSystems Management, SecondEdition

ForthcomingTitles

Smith • LearningOperationsResearchThroughPuzzlesandGames

Tone • AdvancesinDEATheoryandApplications:WithExamplesin ForecastingModels

StochasticModels

Ibe • RandomWalkandDiffusionProcesses

ForthcomingTitles

Donohue,Katok,andLeider • TheHandbookofBehavioralOperations

Matis • AppliedMarkovBasedModellingofRandomProcesses

AHandbookonMulti-Attribute

Decision-MakingMethods

OmidBozorg-Haddad

UniversityofTehran Alborz,Iran

BabakZolghadr-Asli

UniversityofTehran Alborz,Iran

HugoA.Loáiciga

UniversityofCalifornia SantaBarbara,UnitedStates

Thiseditionfirstpublished2021 ©2021JohnWileyandSons,Inc.

Allrightsreserved.Nopartofthispublicationmaybereproduced,storedinaretrievalsystem,or transmitted,inanyformorbyanymeans,electronic,mechanical,photocopying,recordingorotherwise, exceptaspermittedbylaw.Adviceonhowtoobtainpermissiontoreusematerialfromthistitleis availableathttp://www.wiley.com/go/permissions.

TherightofOmidBozorg-Haddad,BabakZolghadr-Asli,andHugoA.Loáicigatobeidentifiedasthe authorsofthisworkhasbeenassertedinaccordancewithlaw.

RegisteredOffice

JohnWiley&Sons,Inc.,111RiverStreet,Hoboken,NJ07030,USA

EditorialOffice 111RiverStreet,Hoboken,NJ07030,USA

Fordetailsofourglobaleditorialoffices,customerservices,andmoreinformationaboutWileyproducts visitusatwww.wiley.com.

Wileyalsopublishesitsbooksinavarietyofelectronicformatsandbyprint-on-demand.Somecontent thatappearsinstandardprintversionsofthisbookmaynotbeavailableinotherformats.

LimitofLiability/DisclaimerofWarranty

Whilethepublisherandauthorshaveusedtheirbesteffortsinpreparingthiswork,theymakeno representationsorwarrantieswithrespecttotheaccuracyorcompletenessofthecontentsofthiswork andspecificallydisclaimallwarranties,includingwithoutlimitationanyimpliedwarrantiesof merchantabilityorfitnessforaparticularpurpose.Nowarrantymaybecreatedorextendedbysales representatives,writtensalesmaterialsorpromotionalstatementsforthiswork.Thefactthatan organization,website,orproductisreferredtointhisworkasacitationand/orpotentialsourceoffurther informationdoesnotmeanthatthepublisherandauthorsendorsetheinformationorservicesthe organization,website,orproductmayprovideorrecommendationsitmaymake.Thisworkissoldwith theunderstandingthatthepublisherisnotengagedinrenderingprofessionalservices.Theadviceand strategiescontainedhereinmaynotbesuitableforyoursituation.Youshouldconsultwithaspecialist whereappropriate.Further,readersshouldbeawarethatwebsiteslistedinthisworkmayhavechangedor disappearedbetweenwhenthisworkwaswrittenandwhenitisread.Neitherthepublishernorauthors shallbeliableforanylossofprofitoranyothercommercialdamages,includingbutnotlimitedtospecial, incidental,consequential,orotherdamages.

LibraryofCongressCataloging-in-PublicationData

Names:Bozorg-Haddad,Omid,1974-author.|Zolghadr-Asli,Babak,author.| Loáiciga,HugoA.,author.

Title:Ahandbookonmulti-attributedecision-makingmethods/Omid Bozorg-Haddad,BabakZolghadr-Asli,HugoA.Loáiciga.

Description:Hoboken,NJ:Wiley,2021.|Series:Wileyseriesin operationsresearchandmanagementscience|Includesbibliographical referencesandindex.

Identifiers:LCCN2020034049(print)|LCCN2020034050(ebook)|ISBN 9781119563495(cloth)|ISBN9781119563471(adobepdf)|ISBN 9781119563389(epub)|ISBN9781119563501(obook)

Subjects:LCSH:Multiplecriteriadecisionmaking.

Classification:LCCT57.95.B692021(print)|LCCT57.95(ebook)|DDC 658.4/03–dc23

LCrecordavailableathttps://lccn.loc.gov/2020034049

LCebookrecordavailableathttps://lccn.loc.gov/2020034050

CoverDesign:Wiley

CoverImage:©connel/Shutterstock

Setin9.5/12.5ptSTIXTwoTextbySPiGlobal,Chennai,India

Contents

Preface xiii

1AnOverviewoftheArtofDecision-making 1

1.1Introduction 1

1.2ClassificationofMADMMethods 5

1.2.1PreferenceEvaluationMechanism 5

1.2.2Attributes’Interactions 7

1.2.3TheMathematicalNatureofAttributes’Values 8

1.2.3.1DeterministicVs.Nondeterministic 8

1.2.3.2FuzzyVs.Crisp 8

1.2.4NumberofInvolvedDecision-makers 8

1.3BriefChronicleofMADMMethods 9

1.4Conclusion 10 References 12

2SimpleWeightingMethods:WeightedSumandWeighted ProductMethods 17

2.1Introduction 17

2.2TheWeightedSumMethod 20

2.2.1Step1:DefiningtheDecision-makingProblem 20

2.2.2Step2:NormalizingtheElementsoftheDecision-matrix 21

2.2.3Step3:AggregatingthePreferenceofAlternatives 21

2.3TheWeightedProductMethod 21

2.4Conclusion 22 References 22

3AnalyticHierarchyProcess(AHP) 25

3.1Introduction 25

3.2TheHierarchicalStructure 27

3.3ThePairwiseComparison 30

3.4Inconsistency 33

3.5QuadrupleAxiomsoftheAHP 35

3.6StepwiseDescriptionoftheAHPMethod 36

3.6.1Step1:DefiningtheDecision-makingProblem 36

3.6.2Step2:PerformingthePairwiseComparisonThroughtheHierarchical Structure 37

3.6.3Step3:EstimatingthePreferenceValueVectors 37

3.6.4Step4:SynthesizingandComputingtheOverallPreferenceValueof Alternatives 38

3.6.5Step5:EvaluatingtheResults’RationalityandSelectingtheBest Alternative 38

3.7Conclusion 39 References 39

4AnalyticNetworkProcess(ANP) 43

4.1Introduction 43

4.2NetworkVs.HierarchyStructure 45

4.3StepwiseInstructiontotheANPMethod 48

4.3.1Step1:DefiningtheDecision-makingProblem 48

4.3.2Step2:ConductingaPairwiseComparisonoftheElementsofthe Decision-makingProblem 49

4.3.3Step3:FormingtheSupermatrix 52

4.3.4Step4:ComputingtheWeightedSupermatrix 53

4.3.5Step5:ComputingtheGlobalPriorityVectorsandChoosingtheMost SuitableAlternative 53

4.4Conclusion 54 References 54

5TheBest–WorstMethod(BWM) 59

5.1Introduction 59

5.2BasicPrinciplesoftheBWM 62

5.3StepwiseDescriptionoftheBWM 63

5.3.1Step1:DefiningtheDecision-MakingProblem 64

5.3.2Step2:DeterminingtheReferenceCriteria 64

5.3.3Step3:PairwiseComparisons 64

5.3.4Step4:ComputingtheOptimalWeights 65

5.3.5Step5:MeasuringtheInconsistencyofDecision-Makers Judgments 66

5.4Conclusion 67 References 67

6TOPSIS 71

6.1Introduction 71

6.2StepwiseDescriptionoftheTOPSISMethod 72

6.2.1Step1:EstablishingtheFormationoftheDecision-making Problem 73

6.2.2Step2:NormalizingtheElementoftheDecision-matrix 73

6.2.3Step3:ComputingtheWeightedNormalizedPreferenceValues 74

6.2.4Step4:DefiningtheReferenceAlternatives 74

6.2.5Step5:CalculationoftheSeparationMeasure 75

6.2.6Step6:ComputingtheRelativeClosenesstotheIdealSolution 76

6.2.7Step7:RankingtheAlternatives 76

6.3ACommonMisinterpretationofTOPSISResults 76

6.4Conclusion 77

References 78

7VIKOR 81

7.1Introduction 81

7.2StepwiseDescriptionoftheVIKORMethod 84

7.2.1Step1:ModelingtheDecision-MakingProblem 84

7.2.2Step2:NormalizingtheElementoftheDecision-Matrix 85

7.2.3Step3:Computethe“GroupSatisfaction”and“IndividualRegret” Parameters 85

7.2.4Step4:ComputingtheVIKORParameter 86

7.2.5Step5:RankingtheAlternatives 86

7.2.6Step6:DeterminingtheCompromiseSolution 86

7.3Conclusion 87

References 88

8ELECTRE 91

8.1Introduction 91

8.2ABriefHistoryoftheELECTREFamilyofMethods 93

8.3ELECTREI 94

8.4ELECTREII 96

8.5ELECTREIII 99

8.6ELECTREIV 104

8.7Conclusion 105

References 106

9PROMETHEE 111

9.1Introduction 111

9.2CommonGroundofthePROMETHEEFamily 112

x Contents

9.2.1Stage1:ConstructionoftheGeneralizedCriteria 113

9.2.2Stage2:MappingtheOutrankRelationontheSetofFeasible Alternatives 116

9.2.3Stage3:EvaluationtheRelationAmongtheFeasibleAlternatives 116

9.3PROMETHEEI 117

9.4PROMETHEEII 118

9.5PROMETHEEIII 119

9.6PROMETHEEIV 120

9.7Conclusion 121

References 121

10SuperiorityandInferiorityRanking(SIR) 125

10.1Introduction 125

10.2FoundationalBasesoftheSIRMethod 126

10.3StepwiseDescriptionoftheSIRMethod 129

10.3.1Step1:EstablishingtheFormationoftheDecision-Making Problem 129

10.3.2Step2:ComputingtheSuperiorityandInferiorityScores 129

10.3.3Step3:FormingtheSuperiorityandInferiorityMatrices 132

10.3.4Step4:SuperiorityandInferiorityFlows 133

10.3.5Step5:RankingtheSetofFeasibleAlternatives 135

10.4Conclusion 136

References 137

11PAPRIKA 139

11.1Introduction 139

11.2StepwiseDescriptionofPAPRIKA 140

11.2.1Step1:DefiningtheDecision-MakingProblem 141

11.2.2Step2:IdentifyingtheNondominatedPairsofAlternative 141

11.2.3Step3:RankingthePairsofNondominatedSolutions 142

11.2.4Step4:CalculatingtheCompleteRankingofAlternatives 144

11.3Conclusion 145

References 146

12GrayRelationalAnalysis 149

12.1Introduction 149

12.2GraySystemTheory:TheFoundationandBasicPrinciples 150

12.3GrayRelationalModeling 151

12.4GrayTheoryinRelationtoMADM 153

12.5Conclusion 155

References 155

AWeightAssignmentApproaches 159

A.1SubjectiveApproach:WeightedLeastSquares 160

A.2ObjectiveApproach:MultiobjectiveProgrammingModel 162

References 164

BABenchmarkExampleandaComparisonbetweenObjective-and Subjective-BasedMADMMethods 167

References 171

Index 173

Preface

Multi-attributedecision-making(MADM)problemsdescribeasituationinwhich thedecision-makersevaluateafinitenumberofpre-definedalternativesthat areexplicitlyknownatthebeginningofthesolutionprocesswithregardtoa setofoftenconflictiveevaluatingcriteria.Itwouldnotbeanexaggerationto statethatalmosteveryone,whetherintheirpersonalorprofessionallife,faces decision-makingproblemsonadailybasis.AlthoughtherearenumerousMADM methodsatthedecision-makers’disposaltocopewithreal-worlddecision-making problemsthereisstilllackingasourcethatcompilesandexplainsthevarious MADMmethodsinaclearandsystematicmannerthatwouldmaketheirunderstanding,comparison,andapplicationstraightforwardforthosewhorequire implementingthesetechniques.MostexistingsourcesthatdealwithMADM methodsaregenerallyfocusedonresultsoftheapplicationsofthesemethods,but overlookbasicandunifyingconcepts.Therefore,thosewhoareeagertogainan overviewofthesemethodsmustendurehardshipsofsearchingthroughvarious sourceswhicharewritteninspecializedformandaredifficulttounderstand.

Thebookathandfillsthecitedknowledgeandeducationalgapanddescribes themostimportantMADMmethods,withanassessmentoftheirperformancein solvingmultipleproblemsencompassingmanyfieldsofinquiryandpracticeina clearandsystematicmanner.Theproposedbookcontains12chaptersplustwo appendices.Chapter1providesanoverviewofthedecision-makinganditsfundamentalconcepts.Eachofchapters2through12isdevotedtoaseparateMADM method.Intotal,some20MADMmethodsarepresentedinthebook.Appendix Idescribesaweightassignmentmethod;AppendixIIcontainsanapplicationof MADMmethods.Thechaptersarearrangedbasedonpedagogicalpurposessothat theaudiencecaneasilyengagewiththepresentedmaterialsineachgivenchapter. Nevertheless,thebasicideaistoensurethateachchaptercanstandalonebyprovidingtheaudiencewithabriefdescriptionofthematerialsandmethodsrequired tocovereveryaspectandmathematicalconceptsusedineachgivenmethod.In otherwords,whileahighlyengagedaudiencecangothroughtheentirebookto

xiv Preface

gainadeepunderstandingofMADMmethodsandtheirtheoreticalfoundation, somecanaimataspecificchapterwithoutfacinganydifficultyinunderstandingthematerialspresentedforthereviewedMADMmethods.Inessence,each chapterdescribingaspecificMADMmethod,orinsomecaseafamilyofmethods,startswithabriefliteraturereviewofthemethods’developmentfollowed byadescriptionofitstheoreticalorigins.Thephilosophicalfoundationsofeach methodarediscussedandmappedtothemathematicalframeworkofthespecifiedMADMmethod.Eachchaptercontainsastepwisedescriptionofitspertinent methodthatservesastheguidelineforimplementingthemethodwiththepurpose ofcopingwithreal-worldMADMproblems.

February14,2021 IranandUSA

OmidBozorg-Haddad BabakZolghadr-Asli HugoA.Loáiciga

AnOverviewoftheArtofDecision-making

1.1Introduction

Whatmotivatesonetomakeadecision?Findingtheprecisereasonbehindthese motivationsmightnotbeaseasyasitmightseem.Nevertheless,giventhatthese choicesareshapingtheworldaroundus,itwouldnotbeanexaggerationtoclaim thattheanswertotheaforementionedquestionmayfacilitateunderstandingthe workingsofmanyworldphenomena.Justforamomentconsiderthepossibilityofknowingthemotivationspromptingpersontomakedecisions.Ifthatwere achieved,predictinghumans’behaviorfromthesimpleevery-dayactivitytothe mostsophisticatedsocial,economic,andpoliticalcontextswouldbepossible.

Nowletuschangethescopeofthequestion; howcanonemakeagoodchoice? Thistimewemaybemoresuccessfulinfindingamoreproperanswer.Letustakea momenttoconsiderthedescriptionoftheactofdecision-making.TheOxforddictionarydefinesdecision-makingas“theprocessofdecidingaboutsomethingimportant,especiallyinagroupofpeopleorinanorganization.”Fromapsychological pointofview,however,decision-makingisregardedasthecognitiveprocessresultingintheselectionofabelieforacourseofactionamongseveralalternativepossibilities.Eachdecision-makingprocessproducesafinalchoice,whichmayormay notpromptaction(TzengandHuang2011).Inotherwords,thedecision-making merelyreferstotheactchoosingamongasetofsolutions,ratherthantheproceduralrequirementsofexecutingtheselectedsetofalternatives.

Thedecision-makingprocessisfoundedonafour-stageanalyticalprocedure (VroomandJago1974;Belletal.1988;WeberandCoskunoglu1990;Kleindorferetal.1993).Thefirststageofthedecision-makingprocessisbetterknown asdescriptiveanalyticsorpositiveanalytics,whichisconcernedwithdescribing observedbehaviorsofthestakeholderswhoareinvolvedinthedecision-making process,mainlybylookingatthetheirpastperformanceandunderstandingsuch behavioreitherbymininghistoricaldatasets,and/orlookingforthebehavioral andsocial,psychological,andevenneurologyreasoningmotivationsthatcanbest

AHandbookonMulti-AttributeDecision-MakingMethods, FirstEdition. OmidBozorg-Haddad,BabakZolghadr-Asli,andHugoA.Loáiciga. ©2021JohnWiley&Sons,Inc.Published2021byJohnWiley&Sons,Inc.

1AnOverviewoftheArtofDecision-making

describethecourseofactionsmadebythestockholdersofthedecision-making problem(TzengandHuang2011;SantosandRosati2015).Descriptiveanalysisis outsidethescopeofthisbookinspiteofitspsychologicalnaturebeingpertinent todecision-makingprocess.

Thepredictiveanalyticstageconcernsthepredictionofwhatislikelytooccur givenasetofcircumstances,whichtakesplaceafterdiscerningthemotivational patternsbehinddecision-makingproblemsthroughdescriptiveanalyticaltechniques.Theapplicationofpredictiveanalysisislimitedtothedecision-making underuncertaintyand,admittedly,notalldecision-makingproblemsrequire suchapproach.Nevertheless,ifnecessary,thehistoricaldatasetsmaybereviewed duringthissecondstagetodeterminetheprobabilityofaneventorthelikelihood ofasituation’soccurrence(Belletal.1988;Kleindorferetal.1993).Exploringthis phaseofthedecision-makingprocessislefttoreadersgiventhescopeandaims ofthisbook.

Thethirdstageofthedecision-makingprocessisthenormativeanalysis.The term“normative”generallyreferstorelatinganitemtoanevaluativestandard throughassessingandmakingjudgmentsabouttheitem’sbehaviororoutcomes (KahnemanandTversky1984;TverskyandKahneman1986).Normativeanalysis, subsequently,isconcernedwithtechniquesthroughwhichthedecision-makers wouldbeabletoevaluatethefeasiblealternativesinamathematicalsense(Bell etal.1988;Kleindorferetal.1993).Notethattraditionalnormativeanalysis isbasedontheassumptionofrationalismthroughtheevolvedentitiesofthe decision-makingproblem,which,looselyspeaking,isatermthatrefersto decision-makerspursuingwhatwasdescribedthroughthepreviousstages,as theirinterestsandgoals.Naturally,makingadecisionirrationallyisbeyondthe scopeofthisbook,thoughmethodsareintroducedthroughoutthisbookthat wouldenabledecision-makerstocopewithdifferenttypesofcriteria,including intangiblecriteria.

Thefinalstageofthedecision-makingprocessiscalledprescriptiveanalysis.In thisstage,decision-makersgobeyondpredictingfutureoutcomestodetermine whichalternativeswouldbethemostadvantageousordesirablesolutionsto theproblemathand(Saad2001).Inotherwords,prescriptiveanalyticscombinetheinformationgatheredthroughstudyingthebehavioralpatternsofthe stockholders(descriptiveanalysis),thelikelihoodofrandomeventsinherentto thedecision-makingproblems(predictiveanalysis),whichwouldbeexpressed inmathematical-orientedframeworks(normativeanalysis),toobtainthebest courseofactionsforthedecision-makers.Furthermore,throughtherealms ofprescriptiveanalysis,decision-makerscanexplorethepossibleoptionson howtotakeadvantageoffutureopportunitiesorcopingwithfuturerisks,and, eventually,evaluatetheimplicationofeachfeasibledecisionoptionbasedonthe naturethedecision-makingproblemathand(Belletal.1988;Kleindorferetal. 1993;TzengandHuang2011).

Havingdefineddecision-making,weconsiderwhatisagoodchoiceoralternativeinadecision-makingproblem.Indeed,thenotionofa“goodalternative” maydifferamongdecision-makers’viewpointsduetotheirdifferentpersonal desires,experiences,andbackgrounds.Inotherwords,one’sideaofa“good choice”maynotnecessarilyrepresenteverydecision-makers’idealchoice. Furthermore,theselectionprocedureofdecision-makersmaydifferfromone another,whenfacingthesamedecision-makingproblem.Nevertheless,the decision-makers’selectionprocedureisfoundedonabasicandsimilarprinciple, whichisthatdecision-makerswouldhavetochooseasetofsolutionsthatwould outperformotherfeasiblealternativesbasedonasetofevaluationcriteriadefined eitherexplicitlyorimplicitlybythedecision-makersforthespecificproblemat hand.Infact,thisdecisionparadigmunderliesmulticriteriadecision-making (MCDM)ingeneral.Inpractice,almosteveryonemayfaceanMCDMproblem onadailybases,whichmostcopewithbyaggregatingthecriteriathrough anintuition-orientedweightingmechanism.Neverthelessimplementinga systematicMCDMapproachisessentialtomakinginformedandlogicaldecision.

Intechnicalterms,MCDMisaprocedurebywhichthedecision-makerexplicitly evaluatesasetofalternativeswithregardtomultiple,usuallyconflicting,criteria. DecisionmakersapplyMCDMstorestructureandredefinethedecision-making problemtomakeaninformeddecision.Althoughdevelopingandimplementing MCDMmethodsarenotnovelideas,therehavebeenundeniableadvancesinthis fieldsincethebloomingeraofcomputationalintelligence(CI)duringtheearly 1960and1970s,especiallyintheformofmathematicallyorientedmethodsthat recaptureandredefineMCDM.MCDMhasbeenanactiveareaofresearchthathas playedacrucialroleinanarrayofdisciplines,rangingfrompoliticsandbusiness totheenvironmentandenergy(Zolghadr-Aslietal.2018a).

HwangandYoon(1981)proposedclusteringMCDMproblemsbasedonthe natureofsolutionsthatareavailablefortheprobleminhandintotwomain categories,namely,multiobjectivedecision-making(MODM),andmultiattribute decision-making(MADM).Essentially,theaforementionedclassificationis basedonwhetherthesolutionsareexplicitlyorimplicitlydefined(Mendozaand Martins2006;TzengandHuang2011;VelasquezandHester2013).

MODMproblemsdescribeasituationinwhichdecision-makersaresearching forasetofsolutionsthatwouldsatisfytheconstraintsimposedonthegivenproblemandobtainresultsthatconstituteanoptimalsetofsolutionsbasedonthe decision-makersobjectives(HwangandYoon1981).Inessence,MODMissuitablefortacklingdesignandplanningproblems,inwhichthedecision-makersaim toachievestatesobjectivesorgoalsbyconsideringthevariousinteractionswithin thegivenconstraints.ThedecisionspaceofMODMproblemscanbedescribed asamultidimensionalCartesianspace,witheach(conflicting)objectiveacting asanaxis,definedbyasetofconstraintsthatseparatethefeasibleandinfeasible solutions.MODMcansolveproblemswithcontinuousordiscretedecisionspaces.

MODMsolutionmethodsareusuallyassociatedwithmathematicalprogramming methods(TzengandHuang2011).

Ingeneral,MODMinvolvetrade-offandscaleproblems(TzengandHuang 2011;Zolghadr-Aslietal.2018a).MODMinvolvesmorethanoneobjective, therefore,theoptimalsolutionstoaMODMproblemmustbeposedinterms ofParetofrontsorproductionpossibilityfrontier(aftertheItalianeconomist VilfredoPareto1848–1923),whicharesetsofpointsrepresentingcombinationsof thevaluesoftheobjectivefunctionswiththebesttradeoffsamongobjectivesthat areachievablefortheproblembeingsolved.InclassicMODMtechniques,an optimalsolutioniscommonlyobtainedwithmathematicalprogramming.This meansmultipleobjectivesaremergedintoasingle-objectiveproblemthrougha weightingofthevariousobjectives.Theprocessofobtainingaproperweighting schemefortheobjectivesisatrade-offproblem.Ifsuchtrade-offinformation isunavailable,Paretosolutionsmustbederived.ParetosolutionstoMODM problemsareexpressedasasetofnondominatedsolutions.Anondominated solutionhasthepropertythatitisnotpossibletoimprovethesolution’sutility ordegreeofpreferencewithoutdegradingatleastoneobjective(Zolghadr-Asli etal.2017,2018a).TheMODM’sscalingproblem,ontheotherhand,isacomputationalchallengesurroundingmostreal-world,practical,decision-making problems,wherebythestakeholdersmustconsiderseveralconflictingobjectives. Asthenumberofobjectivesincreasesthedecisionmakersface thecurseof dimensionality,wherebythecomputationalcostsofsolvingaMODMproblem becomeburdensomeintheextreme,andsometimescomputationallyunassailable(Bozorg-Haddadetal.2017;Zolghadr-Aslietal.2017,2018a).Inanattempt tosurmountthischallenge,meta-heuristicalgorithmshavearisentosearch withinthedecision-spaceandidentifypotentialsolutionstoaMODMproblem (Bozorg-Haddadetal.2017,Zolghadr-Aslietal.2018b,c,d).

MADMproblemsdescribeasituationinwhichthedecision-makersevaluatea finitenumberofpredefinedalternatives.Thealternativesareknownatthebeginningofthesolutionprocess.Thedecision-makersattempttosystematicallyassess eachalternativeviaadiscretepreferenceratingmechanism.Theratingmechanismusedbydecision-makerstoevaluateandcomparetheperformanceofeach ofthealternativesunderconsiderationisdefinedeitherexplicitlyorimplicitly (HwangandYoon1981).Table1.1comparesthemaincharacteristicofMCDM approaches,namely,MODMandMADM(Malczewski1999;MendozaandMartins2006;TzengandHuang2011;VelasquezandHester2013).

Nowadays,decision-makershavenumerousmethodsandtechniquesat theirdisposaltodealwithMCDMproblems,rangingfromsimple,easy-to-use approachestocomplextechniques.GiventheimportantroleofMADM,itis vitalfordecision-makerstoknowthemeritsanddrawbacksofMADMmethods. ThisbookintroducessomeofthefundamentalMADMmethodsthathavebeen

Table1.1 ComparisonofMODMandMADMapproaches.

CriteriaforcomparisonMODMMADM

CriteriadefinedObjectivesAttributes ObjectivedefinedExplicitlyImplicitly AttributesdefinedImplicitlyExplicitly ConstraintsdefinedExplicitlyImplicitly AlternativesdefinedImplicitlyExplicitly NumberofalternativesInfiniteFinite Decision-makers’controlSignificantLimited DecisionmodelingparadigmProcess-orientedOutcome-oriented RelevanttoDesign/operationEvaluation/choice

proventobeeffectiveandpracticalsolution-searchingtools.Section1.2describes thefoundationsofMADMmethodsandtheirclassification.

1.2ClassificationofMADMMethods

TherearenumerouswaysthroughwhichonecanclassifytheMADMmethods.A soundclassificationreliesonthecoreprinciplesandassumptionsofMADMtocategorizethesemethods.FamiliaritywithMADMmethodsisparamounttochoose adequatelyamongthemtosolveadecisionproblemathand.Accordingly,based onthedecision-makers’prioritizingsystem,theinteractionsamongattributes,the mathematicalnatureofattributes’values,andthenumberofdecision-makers, numerousclassificationshavebeenproposedforMADMmethods.

1.2.1PreferenceEvaluationMechanism

EveryMADMmethodrequiresapreferenceevaluationmechanismforthepurposeofreflectingthestockholders’preferencesinthedecision-makingprocess. Thesemechanismsactasameasurethatenablesdecision-makersevaluating alternativesaccordingtotheirattributes.Themechanismcanbedefinedeither explicitly(wherethepreferencevaluesarecomputablethroughasetofpredefined boundariesormathematicalfunctions),ortheycanbedefinedimplicitlysothat thedecision-makers’experiences,expertise,perception,andinstinctsarereflected inthealternatives’preferenceevaluation.

Aclassificationbasedonthenotionofpreferenceevaluationmechanismdivides MADMmethodsintomultiattributeutilitytheory(MAUT)andoutranking

1AnOverviewoftheArtofDecision-making

methods(BeltonandStewart2002;MendozaandMartins2006).Onthebasis ofBernoulli’sutilitytheoryMAUTmethodsobtainthedecision-makers’preferences,whichcanusuallyberepresentedasahierarchicalstructurebyusingan appropriateutilityfunction.Byevaluatingtheutilityfunctionthealternativewith thehighestutilityvaluecanbeidentifiedasthesolutiontotheMADMproblemat hand.InspiteoftheirrelianceofsolidaxiomaticbackgroundofMAUTmethods, theyarecriticizedbytheirunrealisticassumptionofpreferentialindependencies (TzengandHuang2011).Preferentialindependencedescribessituationsinwhich thepreferredoutcomeofonecriterionoveranotherisnotinfluencedbythe remainingcriteria.However,itsohappensthatthecriteriaareusuallyinteractive inreal-worldMADMproblems.Alternatively,insteadofbuildingcomplex utilityfunctions,outrankingmethodscomparethepreferencerelationsamong alternativestodetermineonthebestalternatives.Theoutrankingmethodswere introducedtoovercometheempiricaldifficultiesexperiencedwiththeutility functioninhandlingpracticalproblems.Yettheylackaxiomaticfoundation,such asisthecasewithclassicalaggregateproblemsandstructuralproblems(Tzeng andHuang2011).

ThepreviousclassificationcategorizesclassicalMADMmethods,yetitmay facedifficultiescategorizingsomeofthemodernMADMmethodswhosefeatures donotfiteitheroneofthepreviouslycitedcategories.BeltonandStewart(2002) proposedamoresophisticatedclassificationsystemforMADMmethodsthat addressesthelatterclassificationdifficulties.ThealternateclassificationcategorizestheMADMmethodswithinthreeclasses,namely,valuemeasurement,goal aspirationsorreferencelevel,andoutrankingmethods.Thisclassificationisa reviewednext.

Value-measurementMADMmethodsimplementnumericalscalestorepresent thedegreetowhichafeasiblealternativemaybepreferabletoanother.Thescores obtainedforeachalternativearedevelopedinitiallyforeachindividualevaluation criterionandarethensynthesizedtoratetheoverallperformanceofthealternatives.Thescoresassignedtoeachofthefeasiblealternativesreflectapreference order.Thesepreferencesmustbeconsistentwithasetofaxioms,whichareas follows(BeltonandStewart2002;MendozaandMartins2006):

(I)Constantdisciplineandrolesmustbeimposedbythedecision-makerinthe constructionprocedureofpreferencemeasurementscales;

(II)Provideaframeworkthroughwhichthedecision-makersareabletosystematicallyanalyzetheobtainedpreferencevaluesandgainadeeperunderstandingoftheprocessthatledtothefinalresults;and

(III)Promotingexplicitstatements,ratherthanimplicitjudgmentsregardingthe trade-offsbetweenevaluationcriteria.

Desirableorsatisfactorylevelsofachievementmustbedefinedbythe decision-makersforeachevaluationcriterion.Throughthesereferencelevel methods,thosealternativesthatareclosesttoachievingthegoalsoraspirations areidentified.ThesetypesofMADMmethodsarerecommendedforthose casesinwhichdecision-makersmaynotbeabletoexpresstrade-offsoridentify importanceweightsoftheevaluationcriteria.Nevertheless,themostdesirable outcomecanbeportrayedthrougharbitraryaspirationsorgoalsforeachcriterion. AsfarasthesebranchesofMADMisconcerned,analternativethatrepresents themostsimilaritieswiththearbitrarydefinedidealsolutioncanbestreflectthe stakeholders’interestsintheprocessofdecision-making.Throughtheframework representedbythisbranchofMADMmethods,thefeasiblealternatives,which areavailablecoursesofactions,aresystematicallyeliminateduntilachievinga solutionthatbestfitsthestakeholders’idealoutcomefortheMADMproblemat hand(BeltonandStewart2002;MendozaandMartins2006).

Lastly,outrankingMADMmethodsevaluatealternatives’relativeperformances againstoneanotherusingacomparison-orientedframework.OutrankingMADM methods,thefirstevaluatefeasiblealternativesbythedecision-makersinterms ofevaluationcriteriatoestablishtheirmerits.Thisisfollowedbyanaggregation stagewherebythegatheredinformationisusedasevidencetoobtainanalternative thatoutrankothersandemergesastheoptimalsolution.Theaggregationstage establishestherelationsbetweenthealternativesintermsofpreference,indifference,andincomparability.Consequently,acompleterankingofalternativesis produced.

1.2.2Attributes’Interactions

InanyMADM,thedecision-makerisdealingwiththepresenceofanumberof evaluationcriteria.Inessence,eachMADMmethodoffersadifferentapproach toaggregateeachcriterion’svaluetoobtainanoptimalsolution.Basedonthat notionMADMmethodsaredividedintotwomaincategories,namely,compensatoryandnoncompensatory(Jeffreys2004).Incompensatorytechniques,the poorperformanceofanalternativeinsomecriteriacanbecompensatedbyhigh performanceinsomeothercriteria.Therefore,theaggregatedperformanceof analternativemightnotrevealitsweaknessarea.Incontrast,innoncompensatorytechniques,thesignificantpoorperformanceofanalternativeinsome criteriacannotbecompensatedwithhighperformanceinothercriteria.The aggregatedperformancereflectthisfact.Inotherwords,eachcriterioncan independentlyplayacrucialruleintheaggregatedperformanceofanalternative (Banihabibetal.2017).

1.2.3TheMathematicalNatureofAttributes’Values

Fromamathematicalpointofview,variables,andinthiscaseevaluatingcriteria, canhavedifferentnature,suchas,deterministicvs.nondeterministic,andfuzzy vs.crisp.MADMmethodscanbedividedintothefollowingcategories.

1.2.3.1DeterministicVs.Nondeterministic

DeterministicMADMmethodsinvolvedecision-makerswhoarecertainabout theoccurrenceofthesetofoutcomesinadecision-makingproblem.Onthe otherhand,nondeterministicproblemsinvolvetheoccurrenceofoutcomeswith stochasticcomponentsofarandom-basednature(Pearl1996;TzengandHuang 2011).Insuchcase,thelikelihoodofanoutcomewouldplayadirectroleinselectingthemostsuitablealternative(CoombsandPruitt1960).Nondeterministic methodsarebeyondthescopeofthisbook.

1.2.3.2FuzzyVs.Crisp

CripMADMmodelingexpressesthedecision-makers’preferenceswithnumeric values.However,therearecasesinwhichthesubjectiveuncertaintiesthataresurroundingdecision-makerspreventthestockholderstoexpresstheirpreferences withacrispnumber(TzengandHuang2011).Insuchsituations,decision-makers mayrelyonafuzzysetthatcanbestdescribethestockholders’preferences.Fuzzy setsofferthebenefitofimplyinglinguisticevaluation,whichinturn,wouldease theevaluationprocessofthedecision-makers(BellmanandZadeh1970).

Itisvitalfordecision-makerstodistinguishthefuzzy-uncertaintylogicfrom theprobability-uncertaintylogic,andtousetheminthepropercontext.Incases wherethecertaintyofoutcomesisinquestion,theprobability-uncertaintylogic istherecommendedtool.Insuchsituations,thedecision-makers’decision-treeis foundedonatleastoneuncertainevent.Consequently,theprobabilityofeachoutcomewouldplayaroleindeterminingthemostsuitablealternative.Ontheother hand,whenthedecision-makersarenotcertainonhowtoexpressthepreference ofanalternative,thefuzzylogicbecomesthefavoredoption.Fuzzyevaluation enablesdecision-makerstodescribeanalternative’spreferencethroughafuzzyset employingmembershipfunctions.Inessence,whiletheprobability-uncertainty logicdealswiththeprobabilityofoutcomesinadecision-tree,thefuzzylogic offersthepossibilityofpreferenceevaluationbythedecision-makers.Exploring therealmsofnondeterministicevaluationandfuzzydescriptionofperformances laysbeyondthescopeofthisbook.

1.2.4NumberofInvolvedDecision-makers

MADMmethodscanbeclassifiedassingleorgroupdecision-makingmethods dependingonthenumberofdecision-makersinvolved(Black1948).Inthe

caseofsingledecision-makermethods,theopinionofthatsingleindividual formsthepreferenceevaluationmechanismofthedecision-makingprocess. Ontheotherhand,groupdecision-makingenablesanumberofexpertsand stakeholderstocontributeandinfluencethedecision-makingprocess(Kiesler andSproull1992).Groupdecision-makingmethodsarefoundedonthebasisof singledecision-makingmethods;yet,theyrequireanadditionalstrategythrough which,eachdecision-maker’sopinionisaggregatedandintegratedwithothers’ viewpointstoformthefinalresult.Exploringsuchstrategiesfallsoutsidethe scopeofthisbook.

1.3BriefChronicleofMADMMethods

ThehistoricaloriginsofMADMcanbetracedbacktoseriesofcorrespondence letterbetweenNicolasBernoulli(1687–1759)andPierreRémonddeMontmort (1678–1719),whilediscussingamathematicalbrainteaser,knownasthe St.Petersburgparadox (TzengandHuang2011).Inbrief,theSt.Petersburgparadoxcanbe portrayedasfollows(Bernstein1996):

“Thisisagameofchanceforasingleplayerwhotossesafaircoinateach stageofthegame.Theplayerkeepstossingthecoinuntilitturnstails.If thefirstflipistailstheplayerwins$2;ifthefirsttailsisonthesecondflip theplayerwins$4;ifthefirsttailsisonthethirdfliptheplayerwins$8,etc. Concretelyiffirsttailsisonthenthfliptheplayerwins$2n .”Thequestion hereis: howmuchwouldaprospectivegamblerbewillingtopaytoplaythis game?

Tograspthemagnitudeofthedescribedconundrum,considerforamoment, theanswerofclassicalmathematicstothedescribedquestion.Theexpectedvalue oftheprizeresultingfromplayingthisgameis(Bernoulli1738):

(1.1)

inwhich EV = theexpectedvalueturnsouttobeinfinity.Accordingly,aplayer wouldbewillingtopayanypricetoparticipateinthedescribedgame.However, thisresultdefieshumanbehaviorsincenoonewouldbewillingtopayalimitless amountofcashtoengageinthisgame(RiegerandWang2006).Theanswertothe St.Petersburgparadox,whichrevolutionizedthewayinwhichdecision-making problemswereanalyzed,didnotsurfaceitselfuntilDanielBernoulli(1700–1782)

1AnOverviewoftheArtofDecision-making

publishedhisinfluentialresearchonutilitytheoryin1738.TheconcretediscussionsdescribingthesolutionoftheSt.Petersburgparadoxindetailareskipped here;yet,itisnoteworthythattheremarkablesolutionthatenabled Daniel Bernoulli tosolvetheaforementionedparadoxreliedonthefactthathumans makedecisionsbasednotontheexpectedvalue,butrather,ontheutilityvalue. Specifically,assumethataprospectiveplayerhasawealthof w dollars,thatthe chargeforenteringthegameequals c dollars,andthattheplayer’sutilityfunction is U (w) = ln(w).Itcanbeshownthatunderthesecircumstances,theexpected incremental(ormarginal)utilityofplayingthisgame[EΔ(U )]isfinite:

Therefore,aprospectiveplayerwhosewealthequalsUS$106 shouldbewillingto payuptoUS$20.88toplaythegame;orUS$10.95ifthewealthisUS$103 ,andso onandsoforth,becausetheamountstheplayerwouldbewillingtopaymaximizehisexpectedincrementalutility.Theimplicationoftheutilityvalueisthat humanschoosethealternativewiththehighestexpectedutilityvaluewhenconfrontingtheMADMproblems.Achronologicoverviewofthemostfundamental andinfluentialMADMmethods,whichwouldbediscussedwithinthisbook,is presentedinTable1.2.

1.4Conclusion

Almosteveryone,onadailybases,facesdecision-makingproblems.Itwouldnot beexaggeratedtostatethatthesedecisionsconstitutethenatureofmankindand ofthesocietythathumansform.Whenitcomestoreal-worlddecision-making problems,thedecision-makersoftenfindjudgmentachallengingtask.This issobecauseofthenotionthattheinterestofthestakeholderscanbeonly representedthroughtheevaluationofasetofconflictivecriteria.Whenever thedecision-makersfaceasetoffeasible,discrete,alternatives,theproblem athandinvolvesMADM.Numerousmethodshavebeenpresentedbybeen reportedtoensureasoundandreliabledecision-makingprocess.MADMisone ofthemainbranchesofoperationalresearch;itisanactivefieldofstudywith multipleoverlapswithmanyscientificdisciplines,andhasnumerouspractical applications.ThischapterreviewedtheprinciplesofMADM.Furthermore,the bestwell-knownMADMswerehereinclassifiedandreviewed.

Table1.2 AchronologicoverviewofthemostinfluentialMADMmethods.

MADMMethodsUtilityfunctionBernoulli(1738)

Weightedsummethod (WSM) ChurchmanandAckoff(1954)

ELECTEREIBenayounetal.(1966)

ELECTEREIIRoyandBertier(1971)

Analytichierarchy process(AHP) Saaty(1977)

ELECTEREIIIRoy(1978)

TOPSISHwangandYoon(1981)

ELECTEREIVRoyandHugonnard(1982)

PROMETHEEIBrans(1982)

PROMETHEEIIVinckeandBrans(1985)

PROMETHEEIIIBransetal.(1986)

PROMETHEEIVMladineoetal.(1987)

GreyrelationalanalysisDeng(1989)

Analyticnetwork process(ANP) Saaty(1996)

VIKOROpricovic(1998)

Superiorityand inferiorityranking (SIR) Xu(2001)

PAPRIKAHansenandOmbler(2008)

Best-worstmethod (BWM) Rezaeietal.(2015)

WeightingMethodsEntropymethodShannon(1948)

DelphimethodDalkeyandHelmer(1963)

EigenvectormethodSaaty(1977)

Weightedleastsquare method Chuetal.(1979)

Multipleobjective programmingmodel ChooandWedley(1985)

Principalelement analysis Fan(1996)

ModifiedDelphi method Custeretal.(1999)

References

Banihabib,M.E.,Hashemi-Madani,F.S.,andForghani,A.(2017).Comparisonof compensatoryandnon-compensatorymulticriteriadecisionmakingmodelsin waterresourcesstrategicmanagement. WaterResourcesManagement 31(12): 3745–3759.

Bell,D.E.,Raiffa,H.,andTversky,A.(1988). DecisionMaking:Descriptive,Normative, andPrescriptiveInteractions.Cambridge,UK:CambridgeUniversityPress.

Bellman,R.E.andZadeh,L.A.(1970).Decision-makinginafuzzyenvironment. ManagementScience 17(4):141–164.

Belton,V.andStewart,T.(2002). MultipleCriteriaDecisionAnalysis:AnIntegrated Approach.Massachusetts,BST:KluwerAcademicPublishers.

Benayoun,R.,Roy,B.,andSussman,B.(1966).ELECTRE:Uneméthodepourguider lechoixenprésencedepointsdevuemultiples.Notedetravail49,SEMA-METRA International,DirectionScientifique,Paris,France.

Bernstein,P.(1996). AgainsttheGods:TheRemarkableStoryofRisk.NewYork,NY: Wiley.

Bernoulli,D.(1738).Specimentheoriaenovaedemensurasortis. Comentarii AcademiaeScientiarumImperialesPetropolitanae 1738(5):175–192.

Black,D.(1948).Ontherationaleofgroupdecision-making. JournalofPolitical Economy 56(1):23–34.

Bozorg-Haddad,O.,Solgi,M.,andLoáiciga,H.A.(2017). Meta-heuristicand EvolutionaryAlgorithmsforEngineeringOptimization.Hoboken,NJ:Wiley.

Brans,J.P.(1982).L’ingénieriedeladecision.Elaborationd’instrumentsd’aideala decision:MethodePROMETHEE.In: L’aidealaDecision:Nature,Instrumentset PerspectivesD’avenir (eds.R.NadeauandM.Landry),183–214.Québec,Canada: PressesdeUniversiteLaval.

Brans,J.P.,Vincke,P.,andMareschal,B.(1986).Howtoselectandhowtorank projects:thePROMETHEEmethod. EuropeanJournalofOperationalResearch 24 (2):228–238.

Choo,E.U.andWedley,W.C.(1985).Optimalcriterionweightsinrepetitive multicriteriadecision-making. JournaloftheOperationalResearchSociety 36(11): 983–992.

Chu,A.T.W.,Kalaba,R.E.,andSpingarn,K.(1979).Acomparisonoftwomethodsfor determiningtheweightsofbelongingtofuzzysets. JournalofOptimizationTheory andApplications 27(4):531–538.

Churchman,C.W.andAckoff,R.L.(1954).Anapproximatemeasureofvalue. Journal oftheOperationsResearchSocietyofAmerica 2(2):172–187.

Coombs,C.H.andPruitt,D.G.(1960).Componentsofriskindecisionmaking: probabilityandvariancepreferences. JournalofExperimentalPsychology 60(5):265.

Custer,R.L.,Scarcella,J.A.,andStewart,B.R.(1999).ThemodifiedDelphi technique:arotationalmodification. JournalofCareerandTechnicalEducation 15(2):50–58.

Dalkey,N.andHelmer,O.(1963).AnexperimentalapplicationoftheDelphimethod totheuseofexperts. ManagementScience 9(3):458–467.

Deng,J.(1989).Introductiontogreysystemtheory. TheJournalofGreySystem 1(1): 1–24.

Fan,Z.P.(1996).Complicatedmultipleattributedecisionmaking:theoryand applications.Ph.D.Dissertation.NortheasternUniversity.Shenyang,China.

Hansen,P.andOmbler,F.(2008).Anewmethodforscoringadditivemulti-attribute valuemodelsusingpairwiserankingsofalternatives. JournalofMulti-Criteria DecisionAnalysis 15(3–4):87–107.

Hwang,C.L.andYoon,K.(1981).Methodsformultipleattributedecisionmaking.In: MultipleAttributeDecisionMaking:LectureNotesinEconomicsandMathematical Systems (eds.C.L.HwangandK.Yoon),58–191.Heidelberg,Germany:Springer PublicationCompany.

Jeffreys,I.(2004).Theuseofcompensatoryandnon-compensatorymulti-criteria analysisforsmall-scaleforestry. Small-ScaleForestEconomics,Managementand Policy 3(1):99–117.

Kahneman,D.andTversky,A.(1984).Choices,values,andframes. American Psychologist 39(4):341.

Kiesler,S.andSproull,L.(1992).Groupdecisionmakingandcommunication technology. OrganizationalBehaviorandHumanDecisionProcesses 52(1):96–123. Kleindorfer,P.R.,Kunreuther,H.,andSchoemaker,P.J.(1993). DecisionSciences:An IntegrativePerspective.Cambridge,UK:CambridgeUniversityPress. Malczewski,J.(1999). GISandMulticriteriaDecisionAnalysis.NewYork,NY:Wiley. Mendoza,G.A.andMartins,H.(2006).Multi-criteriadecisionanalysisinnatural resourcemanagement:acriticalreviewofmethodsandnewmodellingparadigms. ForestEcologyandManagement 230(1–3):1–22.

Mladineo,N.,Margeta,J.,Brans,J.P.,andMareschal,B.(1987).Multicriteriaranking ofalternativelocationsforsmallscalehydroplants. EuropeanJournalof OperationalResearch 31(2):215–222.

Opricovic,S.(1998).Multicriteriaoptimizationofcivilengineeringsystems.Ph.D. Thesis.FacultyofCivilEngineering.Belgrade,Serbia. Pearl,J.(1996).Decisionmakingunderuncertainty. ACMComputingSurveys 28(1): 89–92.

Rezaei,J.,Wang,J.,andTavasszy,L.(2015).Linkingsupplierdevelopmenttosupplier segmentationusingbest-worstmethod. ExpertSystemswithApplications 42(23): 9152–9164.

Rieger,M.O.andWang,M.(2006).CumulativeprospecttheoryandtheSt.Petersburg paradox. EconomicTheory 28(3):665–679.

Roy,B.(1978).ELECTREIII:Unalgorithmedeclassementfondésurune représentationflouedespréférencesenprésencedecritèresmultiples. Cahiersdu Centred’EtudesdeRechercheOpérationnelle 20(1):3–24.

Roy,B.andBertier,P.(1971).LaméthodeELECTREII:Notedetravail142. SEMA-METRA.MetraInternational.

Roy,B.andHugonnard,J.C.(1982).Rankingofsuburbanlineextensionprojectson theParismetrosystembyamulticriteriamethod. TransportationResearchPartA: General 16(4):301–312.

Saad,G.H.(2001).Strategicperformanceevaluation:descriptiveandprescriptive analysis. IndustrialManagementandDataSystems 101(8):390–399.

Saaty,T.L.(1977).Ascalingmethodforprioritiesinhierarchicalstructures. Journal ofMathematicalPsychology 15(3):234–281.

Saaty,T.L.(1996). DecisionMakingwithDependenceandFeedback:TheAnalytic NetworkProcess.Pittsburgh,PA:RWSPublications.

Santos,L.R.andRosati,A.G.(2015).Theevolutionaryrootsofhumandecision making. AnnualReviewofPsychology 66:321–347.

Shannon,C.E.(1948).Amathematicaltheoryofcommunication. BellSystem TechnicalJournal 27(3):379–423.

Tversky,A.andKahneman,D.(1986).Rationalchoiceandtheframingofdecisions. JournalofBusiness:S251–S278.

Tzeng,G.H.andHuang,J.J.(2011). MultipleAttributeDecisionMaking:Methodsand Applications.BocaRaton,FL:CRCPress.

Velasquez,M.andHester,P.T.(2013).Ananalysisofmulti-criteriadecisionmaking methods. InternationalJournalofOperationsResearch 10(2):56–66.

Vincke,J.P.andBrans,P.(1985).Apreferencerankingorganizationmethod:the PROMETHEEmethodforMCDM. ManagementScience 31(6):647–656.

Vroom,V.H.andJago,A.G.(1974).Decisionmakingasasocialprocess:normative anddescriptivemodelsofleaderbehavior. DecisionSciences 5(4):743–769.

Weber,E.U.andCoskunoglu,O.(1990).Descriptiveandprescriptivemodelsof decision-making:implicationsforthedevelopmentofdecisionaids. IEEE TransactionsonSystems,Man,andCybernetics 20(2):310–317. Xu,X.(2001).TheSIRmethod:asuperiorityandinferiorityrankingmethodfor multiplecriteriadecisionmaking. EuropeanJournalofOperationalResearch 131 (3):587–602.

Zolghadr-Asli,B.,Bozorg-Haddad,O.,andChu,X.(2018a).Chapter1:Introduction. In: AdvancedOptimizationbyNature-InspiredAlgorithms.Singapore:Springer InternationalPublishingAG.

Zolghadr-Asli,B.,Bozorg-Haddad,O.,andChu,X.(2018b).Crowsearchalgorithm (CSA).In: AdvancedOptimizationbyNature-InspiredAlgorithms.Singapore: Springer.

Zolghadr-Asli,B.,Bozorg-Haddad,O.,andChu,X.(2018c).Dragonflyalgorithm (DA).In: AdvancedOptimizationbyNature-InspiredAlgorithms.Singapore: Springer.

Zolghadr-Asli,B.,Bozorg-Haddad,O.,andChu,X.(2018d).Krillherdalgorithm (KHA).In: AdvancedOptimizationbyNature-InspiredAlgorithms.Singapore: Springer.

Zolghadr-Asli,B.,Bozorg-Haddad,O.,andLoáiciga,H.A.(2017).Discussionof ‘OptimizationofPhenolRemovalUsingTi/PbO2AnodewithResponseSurface Methodology’byC.García-Gómez,JAVidales-Contreras,J.Nápoles-Armenta,and P.Gortáres-Moroyoqui. JournalofEnvironmentalEngineering 143(9):07017001.

SimpleWeightingMethods:WeightedSumandWeighted ProductMethods

2.1Introduction

Toreachabetterunderstandingofanydecision-makingproblem,onemust employinformation-gatheringmethods,includingbutnotlimitedtosurveys, questionnaires,examination,andsampling,tocollectasmuchpracticalinformationaspossible.Eventually,suchattemptsincreasethechanceofchoosingthe mostsuitablealternative,thatwouldbetterreflecttheneedsandinterestsofthe stakeholdersoftheMADMproblemathand(TzengandHuang2011).

FromtheMADMpointofview,thegatheredinformationregardingtheproblem inquestionisgenerallyrepresentedinamatrixform,commonlyreferredtoasthe decision-matrix.Basedonthedecision-matrix,thedecision-makercananticipate thestakeholders’desiresandpreferences,whicheventuallyleadtochoosingthe mostsuitableavailableoptionthroughamathematicallysupportedframework. Thechoosingprocessproceedsandtheirassumptionsarewhatdistinguishes betweentheMADMmethods.

AMADMproblemiscomposedofasetofalternatives,whicharethefeasible discretesolutionsavailabletothedecision-maker,andasetofevaluationcriteria, whicharetheinstrumentsthroughwhichthestakeholdersdescribetheirobjective.Subsequently,adecision-matrixinextendedformisconstructedbasedonthe fourfollowinginformationsets(Yu1990):

(1)Thesetoffeasiblealternatives,denotedby{ai | i = 1,2, , m}.Noticethateach alternativerepresentsarowinthedecision-matrix(D);

(2)Thesetofpredefinedevaluationcriteriadenotedby{cj | j = 1,2, …, n}.Each criterionrepresentsacolumninthedecision-matrix(D);

(3)Theanticipatedvalueorperformanceofthealternativeswithregardtoeach givencriterion.Let v(i,j) representthevalueofthe ithalternativewithrespect tothe jthcriterion,thena m × n matrixisconstructedwith v(i,j) astheelements;and

AHandbookonMulti-AttributeDecision-MakingMethods, FirstEdition. OmidBozorg-Haddad,BabakZolghadr-Asli,andHugoA.Loáiciga. ©2021JohnWiley&Sons,Inc.Published2021byJohnWiley&Sons,Inc.

2SimpleWeightingMethods:WeightedSumandWeightedProductMethods

(4)Thedecision-makerprioritizesbasedontheweights,denotedby{wj | j = 1,2, …, n}.Each wj reflectstheimportanceofthe ithcriterion.Thisstep involvesaweightingprocedure.

Consequently,thedecision-matrix(D)isrepresentedasfollows(Yu1990):

Inadditiontothedecision-matrixonoccasion,thedecision-makersdefine extremealternatives,namely,ideal(a+ )andinferior(a )alternatives.Theideal alternativeisanarbitrarilydefinedvectorofchoicesdescribingtheaspired solutiontothegivenproblem,which,inpractice,mayormaynotbeachievable. Theinferioralternativeisasolutionthatrepresentsthemostundesirableoption forthegivenMADMproblem.Therearetwomainmethodstocomposethe idealandinferioralternatives.Onecanusethebestandworstvaluesinthe jth columnofthedecision-matrixtocomposethe jthcomponentoftheidealand inferioralternatives,respectively.Ontheotherhand,onecouldalsousetheupper andlowerboundariesofthefeasiblerangeofthe jthcriteriontocomposethese arbitrarilydefinedalternatives.Insuchcases,ifthecriterionisconsideredtobe positive,wherethelargerthevaluethebetterthesituation,theupperandlower boundariesrepresenttheidealandinferioralternatives,respectively.Conversely, anegativecriterion,wherethesmallerthevalue,thebetterthesituation,thelower andupperboundariesrepresenttheidealandinferioralternatives,respectively. Thesearbitrarilydefinedalternativescanthenberepresentedasfollows(Tzeng andHuang2011):

inwhich v+ j and vj = thecomponentsoftheidealandinferioralternativeswith regardtothe jthcriterion,respectively.

TheadmissibilityofeachalternativeinaMADMproblemhingesontheirperformanceswithregardstothepredefinedevaluationcriteria,whichmaybeof differentmathematicalnature.Infact,anMADMproblemcommonlyinvolves multiplecriteriawithdifferentdimensionsandmeasureofscales.Oneofthemain challengesoftheMADMisforthedecision-makertoaggregatetheperformance ofalternativeswithregardtoeachgivencriterionsothattheoverallpreference ofalternativescanbeachieved.However,theformercannotbeachievedwhile theevaluationcriteriaarenotofthesamedimension,measuringunit,andscale. Consequently,throughamathematicalprocedure,betterknownasnormalization,

thedecision-matrixistransformedintoadimensionlessmatrix.Therearevarious normalizationprocedures,suchasthe Z -scoretransformation;yet,thefollowing twoformsarethemostrecommendedforMADMproblems,mainly,becausethey areeasytointerpret(EbertandWelsch2004;Zhouetal.2006;TzengandHuang 2011):

Form I:Thisnormalizationprocess,linearly,transformsalltheperformance values,sothattherelativeorderofmagnitudeoftheratingsremainsequal.The procedurecanbesetupasfollows(ChangandYeh2001):

● Forpositivecriteria:

● Fornegativecriteria:

inwhich r (i,j) = thenormalizedperformancevalueforthe ithalternativeswith respecttothe jthcriterion.

Form II:Inthisnormalizingprocedure,whichisslightlymoreadvancedthan theformertechnique,bothidealandinferioralternativesareusedtonormalize theperformancevalues,asfollows(Maetal.1999):

● Forpositivecriteria:

● Fornegativecriteria:

Thecitednormalizationprocedureyieldsdimensionlessperformancevaluesof thedecision-matrixinwhichthe[r (i,j) ]rangebetween0and1.

Throughthenormalizationprocedure,thedecision-makertransformstheelementsofadecision-matrixintocommensurablevalues.Thenextstepisforthe decision-makertocombinethesevaluesinawaythatthealternatives’overall preferencescanbeevaluated.Herein,assumethatthedecision-makerevaluated theimportanceofeachcriterionandderivethesetofweightsthatbestreflectthe

Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.