搜索
TheStudyonthePatternsofBookingCurvesofTRA立理論林李林年六理Revenuemanagement旅理都歷料利料Bookingcurve落羅列車旅路列車Bookinglimit92年324420料離旅行落例累例念旅車行列車落類數羅落列車落旅量旅異旅例旅連旅不例落落落落列車聯利羅落類類不率不旅落旅羅落旅量(MAPE)說精列車旅量量ABSTRACTThisstudyusesthehistoricaldataofadvancedbookingfromTaiwanRailwayAdministration(TRA)toplotbookingcurves.Severalmethodsincludingclusteranalysis,discriminantanalysis,andmultiplelogisticregressionwereusedtoanalyzethepatternsandcharacteristicsofthebookingcurve,andtoforecasttherailwaypassengertraveldemand.TRAoffersseveralkindsofticketsaleincludinginternetticketingservice,voiceticketingservice,windowadvancedbookingandinstantticketingsales.Afixedquotaofseats(Bookinglimit)oneachscheduleandeachODisset.Thisstudycollectedticketingdataofthehighestclasstrain(Tzu-ChiangTrain)inTRAfrom24thofMarchto20thofAprilin2004,anddivideditintolongandmediumtravelmarketsbythedistance.Theriskpreferenceisusedtoexplainthepassengerticketingbehaviorbygatheringthesimilartrendandloadingratiocurvesintothesamegroupviaclusteranalysis.Then,theexplanatoryvariablesoftheclustersfromtrainsrelatedcharacteristicswereselectedtodiscriminatetheclustersviadiscriminantanalysisandmultiplelogisticregression,andtoexplainandforecasttheclustersattributions,patternsandthepassengerqualityofbookingratiocurves.Theresultsshowthatbookingcurvepatternsoflongtravelmarketsisobviouslydifferentfromthatofmediumtravelmarkets.Thecurvepatternsoflong-termtravelmarketpresentvariegation,peopleusuallypurchaseticketsinadvance;itformsaconservativepurchasingtypetoavoidrisk.Inmedium-termtravelmarket,situationsofadvancedticketingarefew,peopleusuallypurchaseticketsnowadays.Itmeanseveryclusterpresentsarisk-takingpattern.Wecanfindthatclustersindeedgetcorrelationwithtrainstimingcharacteristicsbytheresultsofclusteranalysis.Thepredictionaccuracyofclusterpatterndiscriminationalsogetsprettygoodresultsbyusingdiscriminantanalysisandmultiplelogisticregression.Thepredictionaccuracyofmedium-termstravelmarketsishigherthanthatoflong-termtravelmarkets,andmultiplelogisticregressiongetsbetterpredictionaccuracythandiscriminantanalysisdoes.Besides,whentheaveragevaluesofbookingcurvesfromeveryclusterareusedtoforecastthetraveldemand,themeanabsolutepercentageerror(MAPE)islarge;itmeansanadvancedforecastingmethodsshouldbedevelopedtogetamoreaccuratepassengertraveldemandofTRA.KeyWords:RailwayTransportation,BookingCurve,CusterAnalysis,DiscriminantAnalysis,multiplelogisticregression.論利李林不論念度論老更邏不練老不精神更理立良論論都精論了更論論諸念論更裡LAB裡LAB501樂兩年更AOESC論年來不勵路精神兩年樂論力論林2004年6I錄論---------------------------------------------------------------------------------11.1-----------------------------------------------------------------------------11.2-----------------------------------------------------------------------------31.3-----------------------------------------------------------------------------41.4--------------------------------------------------------------------41.5流--------------------------------------------------------------------5---------------------------------------------------------------------------72.1旅-----------------------------------------------------------72.2----------------------------------------------------------102.2.1----------------------------------------102.2.2羅羅-----------------11論--------------------------------------------------------------------183.1落----------------------------------------------------------------------------183.1.1落-------------------------------------------------------------183.1.2落------------------------------------------------------183.1.3落數量----------------------------------------------------------213.2----------------------------------------------------------------------------223.2.1------------------------------------------------------------223.2.2數-----------------------------------------------------------233.3羅----------------------------------------------------------------243.3.1羅--------------------------------------------------243.3.2羅----------------------------------------25II料-----------------------------------------------284.1----------------------------------------------------------------284.2料----------------------------------------------------------------324.3-------------------------------------------------------374.3.1列車----------------------------------374.3.2北列車----------------------------------39累立-----------------------------------------------------------425.15.1旅累例立---------------------------------------------425.2落----------------------------------------------------------------------------465.2.1落數-------------------------------------------------------------485.2.2落----------------------------------------------------------535.2.3落-------------------------------------------------------------755.3----------------------------------------------------------------------------775.3.1數----------------------------------------------------------775.3.2-------------------------------------------------------------775.3.3-------------------------------------------------------------965