2. Measuring and Predicting Activity-Travel Behavi

MEASURINGANDPREDICTINGADAPTATIONINMULTIDIMENSIONALACTIVITY-TRAVELPATTERNSPROEFSCHRIFTterverkrijgingvandegraadvandoctoraandeTechnischeUniversiteitEindhoven,opgezagvandeRectorMagnificus,prof.dr.R.A.vanSanten,vooreencommissieaangewezendoorhetCollegevoorPromotiesinhetopenbaarteverdedigenopdonderdag8januari2004om16.00uurdoorChang-HyeonJohgeborenteSeoul,KoreaDitproefschriftisgoedgekeurddoordepromotoren:prof.dr.H.J.P.Timmermansenprof.dr.W.W.ReckerCopromotor:dr.T.A.ArentzeCopyright2004C.H.JohTechnischeUniversiteitEindhovenFaculteitBouwkunde,CapaciteitsgroepStedebouwPrintedbytheEindhovenUniversityPressFacilitiesCoverdesign:TonvanGennip,TekenstudioFaculteitBouwkundeCoverillustration:CityofEindhovenCIP-DATAKONINKLIJKEBIBLIOTHEEK,DENHAAGISBN90-6814-577-0ContentsPrefaceListoffiguresListoftables1.Introduction12.MeasuringandPredictingActivity-TravelBehavior:ALiteratureReview52.1Introduction52.2Measuringactivity-travelbehavior62.2.1Measurementpurposes62.2.2Keymeasurementfacets62.2.3Measurementmethods72.3Predictingactivity-travelbehavior102.3.1Comprehensivemodelsoftransportationdemandforecasting112.3.2Studiesonthedynamicsoftravelbehavioranddecisionmaking162.4Conclusionsanddiscussion20PartI.MeasuringActivityReschedulingBehavior3.Uni-dimensionalSequenceAlignmentMethods253.1Introduction253.2Sequencealignmentmethod253.3Anextensionofthealignmentmethod323.3.1Problemdefinition323.3.2Conceptualization333.3.3Aposition-sensitivemethod363.4Illustration443.4.1Ananalysisofsequencesusingtheoriginalsequencealignmentmethod453.4.2Ananalysisofsequencesusingtheextendedsequencealignmentmethod533.5Conclusionsanddiscussion57Appendix3.1:AmathematicalmethodoftheminimumsumofreorderingdistancesAppendix3.2:Appropriatenessproof4.AMultidimensionalSequenceAlignmentMethod674.1Introduction674.2Implicationofmultidimensionalextensionofuni-dimensionalmethod684.3Method694.3.1Problemdefinition694.3.2Conceptualization704.3.3Specificationofthemethod714.4Algorithms784.4.1Aheuristicbasedongeneticalgorithms784.4.2Aheuristicbasedondynamicprogramming844.4.3Ahybridalgorithm874.5Illustration884.5.1FundamentalpropertiesofthemethodexaminedbytheOT-MDSAM884.5.2Performanceofalternativealgorithms914.6.Comparativeanalysisofmeasurementmethods954.6.1Introduction954.6.2Empiricalanalysis954.6.3Conclusions1014.7Conclusionsanddiscussion102Appendix4.1:AnefficientalgorithmfortrajectorysearchandproofofitsoptimalityAppendix4.2:MultiplealignmentsandamultidimensionalalignmentPartII.PredictingActivityReschedulingBehavior5.AModelofActivityRescheduling–AuroraAuroraAuroraAurora1135.1Introduction1135.2Conceptualization1155.3Utilityfunction1195.3.1Basicformoftheactivityutilityfunction1195.3.2Anextensionoftheactivityutilityfunction1245.3.3Totalutilityofrescheduling1275.3.4Adaptationstylesunderuncertainty1285.4Searchtree1305.4.1Theoryandassumptions1305.4.2Model1315.5Numericalsimulations1385.6Conclusionsanddiscussion1486.EstimationMethod1516.1Introduction1516.2Theory1526.3Algorithm1586.3.1Representationofthesolutioncandidates1586.3.2Geneticoperators1596.3.3Geneticparameters1596.3.4Overallprocedure1606.4Modeltest1606.4.1Estimationresultsforsimulatedactivityschedulesusingexactdata1616.4.2Estimationresultsforsimulatedactivityschedulesusingnoisydata1646.5Conclusionsanddiscussion1697.EstimationofAuroraAuroraAuroraAurorausingActivity-DiaryData1737.1Introduction1737.2Theactivity-traveldiarydata1737.3Activity-traveldiarydatausedtoestimatetheutilityfunctions1757.4Estimationofactivityutilityfunctions1797.4.1Principlesandschemeofanalysis1797.4.2Estimationmethod1807.4.3Aggregatemodel1817.4.4Segment-levelmodel1877.5Conclusionsanddiscussion1998.ConclusionsandDiscussion203Bibliography207Authorindex219Subjectindex221EnglishSummary225DutchSummary229KoreanSummary233Curriculumvitae237PrefaceThisbookdescribestheresultsofmyresearchonactivityanalysisoverthelasteightyears.Thecontentcoverstheissuesofmodelingactivitypatternmeasurementandprediction.Thetheoreticalbackgroundoftheproposedmodelforactivitypatternmeasurementisthetime-geographicnotionofinterdependencyandsequentialconnectivitythatreflectsthechoreographyofeverydaylife.Themodelemploysandextendsbiologicalsequentialalignmentmethodstoaddressthemultidimensionalcharacterofactivitypatterns.Thetheoreticalbackgroundofthepredictionmodelstemsfrommicro-economicutilitytheory,time-geographictheoryonpossibilityandconstraints,andpsychologicaldecisionmakingtheory.Themodelpredictsactivitypatternsasaresultofasetofdecisionheuristicsandactivityutilityfunctions.Themeasurementandpredictionmodelarecomplementaryinthesensethatthemeasurementmodelcanbeusedtoidentifysegmentsofhomogeneousactivitypatternssothatthepredictionmodelcanbeestimatedforeachsegmentseparately.Themeasurementmodelcouldpossiblyalsobeusedtomeasurethegoodness-of-fitbetweenpredictedandobservedactivitypatternsinaparameterestimationprocess.TheresearchwasconductedinthecontextoftheAmadeusprojectthatseeksinsightintohowvariouscharacteristicsofmul