呼叫中心排班优化问题建模及研究

中国科学技术大学硕士学位论文呼叫中心排班优化问题建模及研究姓名：张飞飞申请学位级别：硕士专业：金融工程指导教师：方兆本20100408AbstractABSTRACTThroughtheelectronicprocess,callcenterhasextensiveuseinfinancialinstitution.Reasonablescheduleinfinancialinstitutions’CallCenter,notonlycanincreasetherateofcapitalutilization,controlcosts,butalsoimprovetheimageofcorporateservices.HowtooptimizationtheCallCenterSchedulenaturallybecomeoneofthemostimportantquestionsinthefinancialelectronic.Inordertooptimizethecallcenterschedulingsystematically,weneedtorefinetheproblemsintotwoaspects:telephonetrafficandoperatorservicecapabilities.Thispapermakesin-depthresearchonthetelephonetrafficandoperatorservicecapabilities,andintegratesthetwoaspectsintocallcenterschedulingoptimization.Intheprocessofoptimizingschedule,therearemanyeconometricproblems,suchas:extremevalueproblems,periodicproblems.Therefore,areasonableandappliedtelephonetrafficforecastingmodelshouldincludetheperiodic,continuebusy,callback,andextremeproblems.Inthispaperweuselog-linearmodelandrevisedlog-linearmodelofextreme-valuedistributiontoforecasttelephonetraffic,andthenwefindthattherevisedlog-linearmodelofextreme-valuedistributioncandobetterthanlog-linearmodelundertheextremeproblems.Additionally,insystemdesignofcallcenterschedulingoptimization,thepapershowstheintegrationofavarietyoftechnicalmeans.Startfromthecombingofthebasicdata,weobtaindynamicadjustmentofstaffingthroughforecastingtelephonetrafficandoperatorservice.Inthespecificcontentofthearticle,weanalyzethetechnicalaspects,givetheindustrysolutionandputforwardnewideasforthedomesticcallcenterscheduling.KeyWordsCallCenterLog-LinearModelExtremeValueTheoryPossessionCallVolume__________________________________________________________________________________________1111.1CallCenterInternetCTIBICRMPBXERPIT1.1.1199920012002200540020052008122007zIVRzzTGAIVRIP1.1.213CTI/ACD80%1.1.330CallCenterCTIIVRIVRCSRIVR14/1.2zzzzzz1.31.8%2.=P15P=1P=0P=0.5P013.1.4/2622.12.1.1InmonInmonzzzz2.1.227InmonKimball.InmonETLODSInmonInmonInmon2.1InmonRalphKimballFigure3TheDataWarehouseLifecycleToolkit27KimballInmonInmonKimball282.2KimballOLAPinmon2.3292.1.3zzINTERNETz/zzzzsynopses2.42102.2Jenkinson1955GEVGeneralizedextremevaluedistribution()()()()()1exp10exp0xxeHxeeeee--÷ç÷ç-+÷ç÷÷ç-=ìïïï=íïïïîx10xe+e0eFrechet0e=Weibull0eGumbelms()Hxe,,xHHemsems-÷ç=÷ç÷çBMMBlockMaximaModelPOTPeakOverThresholdBMMBlockMaximaPOTthresholdPOT2.32.3.112211345S2.3.2LogisticsLogisticLogistic1Z2ZmZQPQ=lQ=1PQ=0Q=0PQ=l+PQ=0=1()()01122PQllogPQ0mmZZZaaaa==+++?=LogisticSASLOGISTIC212LogisticLogisticlabeledbagofinstancesLogisticLogisticLogistic2.431333.13.1zz314z0709070809z2.3.12.4zRsquare8%3.3.13.2logistic2.3.220087120094303.3R3.3.13.3.1.1315z3.2z3.2Wz3.3z3.23.3.1.2316=-P¾313122829303131256202125262021313.3¾1212128103.4¾200712009125775125313173.5¾tt-1t-2t-1nt-2t-2t3.631808118%¾3.3.1.1P=1P=0P=0.5P01¾PP0.5P¾¾73.3.1.33.3.1.2¾¾0708093.73.83190102030401000001500002000002500001:prdt.ndayreal3.7P=0200812210102030401000001500002000002500001:prdt.ndayreal3.8P=020081221¾3.73.82009112009116¾¾122111111163203.3.1.4Rsquare10.80Rsquare0.8661P=0(100%)P=1()0200400600100000200000Indexa0200400600-0.20.00.2Indexb/a-13.98%321050100150200250100000200000Indexa050100150200250-0.20.00.2Indexb/a-13.1020091203.11-241-251-261-271-281-291-301905091857921639461525811481271498991545742358972166731825281643751556461543541572752185852100411907961786341655651671691689263.21-241-251-261-271-281-291-301542351500641386741342531320311337601433231627521546421506301512941550361504861488851733561663261600511583001413071461151516343.2.2322logisticLogit=++++zLogit=log/1-z=z=logz=z=logz=/20087120094303.3EstimateStd.ErrorzvaluePr(|z|)Sig.(Intercept)1.508e+014.296e-02351.012e-16***log(zhanyong)-1.910e+008.035e-03-237.742e-16***people1.392e-015.062e-04275.022e-16***average.zhanyong-4.930e-033.547e-04-13.902e-16***log(zhanyong):people-1.255e-024.653e-05-269.802e-16***3.113233.112008920091220091200913.1202000400060000500015000Indexaa0200040006000-200001000Indexbb-aa3.12243.133243.13200953.143.14200958%3253.1513.151000013003.153.153.16501100200003.1650003.163263.162707042744.14.14.24.2.11)2)428¾¾¾¾¾3)¾¾¾¾¾¾4.2.24294.2¾¾¾¾4304.34.3.14.3.2¾252425¾32542344.3.31)¾¾2)¾¾3)431¾¾4)¾¾5)¾¾4.3.24324.31)4.42)4334.53)4.64)154344.75)4.84.44.4.14354.4.2¾¾¾¾¾¾4.4.31)2)3)4.4.21)436¾¾¾¾¾2)¾ERD¾¾I/O3)4)ODSODS5)//ODS/4374.94.104.54.115385-0.58%707039[1]955982008652-55[2]2009365-68[3]2006194[4]200829527-30[5]20062243112-114[6]200627214150-4160[7][J]200055(3)24-26[8][J]2004(5)29-32[9][J]200630(10)8284[10]JohnAnton2001[11]VaRES2007(3)118-125[12][M]2005[13][M]1998[14]KimballR.:[M]2003[15]OLAP[J]200623(2)64-67[16][J]2007(3)[17][J]200627(19)3663-3665[18][M]2001487-497[19]AthanassiosN.AvramidisWyeanChanMichelGendreauPierreL’EcuyerOrnellaPisacane.Optimizingdailyagentschedulinginamultiskillcallcenter.EuropeanJournalofOperationalResearch.January2009822-832.[20]LawrenceBROWN,NoahGANS,AvishaiMANDELBAUM,AnatSAKOV,HaipengSHEN,SergeyZELTYN,andLindaZHAO.StatisticalAnalysisofaTelephoneCallCenter:A40Queueing-Science