基于自组织数据挖掘的房地产估价方法及应用研究

基于自组织数据挖掘的房地产估价方法及应用研究管理科学与工程专业研究生谁虹指导教师贺昌政国外房地产估价行业己经有100多年的历史，而我国现代房地产估价的发展才10余年。到目前为止，我国房地产估价方法研究还处于探索阶段。现有的方法注重经验和艺术，对于估价人员本身的素质、从业经验有很高的要求;不同估价师由于主观判断和感受不同，其估价结果差异也较大，且难以判别好坏。另一方面，学术界对于三种传统的房地产估价方法的改进停留在理论研究层面，对于房地产估价的实际操作者来讲，这些方法大多难于理解和运用。因此，如何对目前通用的三种房地产估价的基本方法进行完善与改进，建立一个业内实际操作者认可的、科学的房地产估价方法体系具有重要的理论和实践意义。本文的研究内容主要包括以下四个方面:(l)在国内外房地产估价领域的研究现状和动态分析基础上，通过定性与定量结合的方法，确定房地产价格的影响因素。(2)建立了基于GMDH因果关系检验的房地产估价指标体系，为房地产估价模型打下了良好的基础。d3)在房地产估价传统理论基础上，建立了基于Ac模式识别方法的房地产估价模型。通过实证研究和比较研究，证实了新方法的有效性。(4)将建模理论与估价实务相结合，给出了基于自组织数据挖掘的房地产估价程序，为房地产估价提供了一个可操作性的实现步骤。本文的研究结论与创新主要有以下三点:(l)首次将自组织数据挖掘方法引入房地产估价中，从模式识别的角度建立了基于AC方法的房地产估价模型，给出了基于自组织数据挖掘的房地产估价程序，并通过实证研究证明其科学性和可操作性，扩大了自组织数据挖掘的应用范围，丰富了房地产估价理论与方法体系。(2)与三大传统的方法相比，本文建立的基于自组织数据挖掘的房地产估价方法有效地避免了主观化倾向，并有以下特点:l)基于模式识别的评估方法实质是市场比较法的扩展，其选取的指标更为广泛，采用的样本也更为丰富，减少了在市场比较法中人为选择相似样本、确定相似度的主观性。2)与收益还原法相比，不用人为确定还原利率和预测房地产未来纯收益，评估结果更为客观合理。3)与重置成本法相比，采用房龄作为特征变量之一来替代成本法中房屋折旧的计算，简单易行，更易获得准确的评估结果。4)不仅适合小宗物业的评估，对于大宗物业或者同时对多个物业的评估更加科学、准确和快捷。另外，对于政府进行大规模估价，确定基准价格也有很强的可操作性。(3)与新兴的估价方法相比，本文提出的基于自组织数据挖掘的房地产估价方法有以下两大优势:l)对数据本身的要求不高，样本数量较少的情况下也不影响估价的科学性和准确性;2)方法本身的原理融合了传统估价方法的理论与原则，因此，更加容易被实际估价人员理解和接受，具有较高的可信性和可操作性，且更加优越。关键词:自组织数据挖掘AC模式识别GMDH因果关系检验房地产估价AReseareh&anAPPliedResearehontheRealEstateAPPraisalMethodBasedonSelf-Organ抚ingDataMiningMajor:ManagemeniSeieneeandEngineeringMasterCand1date:QiaoHongSupervisor:HeChangzhengAbstractRealestateaPPraisalintheoverseashasmorethan100yearshistory，butthedeveloPmentofourmodemrealestateaPPraisalonlyhasmorethan10years.Sofar，researehonthetheoryandmethodsofrealestateaPPraisalinoureountry15attheexPlorationstage.ExistingaPPraisalmethodsemPhasizetheexPerienceandart，thusithasaveryhighrequestfortheaPPraiser’5PersonnelqualityandeorrelatedexPerienee.Withdifferenisubjeetivelyjudgesandthedifferentfeelingtothesamerealestate，differentaPPraisers’eanworkoutadifferentaPPraisalresult.Ontheotherhand，researehonthethreetraditionalaPPraisalmethodsintheacademiefieldisattheoryresearehstage，肠theaPPraisaloPerator，theirresearehresultsandimProvementseemdiffieulttounderstandandutilize.Therefore，howtoeonsummate，imProvethePresentgeneralthreerealestateaPPraisalmethodsandestablishseientifie，oPerational，PoPularandquantitativeaPPraisalmethodsystemhasanimPortanttheoryandPraetieesignifieance.ThisartieleresearehcontentsmainlyineludefollowingfourasPeets:First，onthebaseofanalysistothePresentsituationandtendeneyofthedomesticandforeignresearchontherealestateaPPraisal，throughqualitativeandquantitativeanalysismethod，ehoosethemaininfluencefactorsofrealestatePriee.Seeond，establishthemaininflueneefaetorssystemtorealestatePrieebasedontheGMDHeausalitytestingandbuiltthegoodfoundationfortherealestate111aPPraisalmodel.Third，inthefoundationoftraditionaltheoriesoftherealestateaPPraisal，establish“therealestateaPPraisalmodelbasedontheACPattemreeognitionmethod，，，andthroughtheaPPliedresearehandtheeomParisonresearch，thisnewmethod15eonfirnledvalid.Fourth，UnifythemodeltheoryandactualaPPraisalwork，Produee“anoPerationsequeneeaboutrealestateaPPraisalbasedonself-organizingdatamining，’，andProvidefortheaPPraiseraeonereteoPerationalsteP.TheresearehconelusionsandinnovatingPlaceshavefollowing3Points:First，hitrodueedtheself-organizingdataminingmethodtotherealestateaPPraisalforthefirsttime，hasestablishedtherealestateaPPraisalmodelbasedontheself-organizingdataminingfromthePattemreeognitionangleandProdueedanoPerationsequenceforrealestateaPPraisalbasedonself-organizingdatamining，andthroughtheaPPliedresearchhaseonfirmeditseientifieandPossiblyoPerational，hase而ehedtherealestateaPPraisalmethodssystem.Seeond，eomParingwiththreetraditionalaPPraisalmethods，“realestateaPPraisalmethodbasedonself-organizingdatamining，，has4advantages:l)theesseneeofbasedonthePatternreeognitionaPPraisalmethod15exPansionofMarketComP幼sonAPProach.TheseeleetiontargetsaremorewidesPread，thesamPlesarealsorieher，andthismethodcaneffeetivelyredueethesubjectivityinthemarketeomParisonaPProaeh.2)ComParingwithhicomeAPProach，it15notneeessarytodeterminatetheretumtooriginalestateinierestrateandtoforeeastthefuturenet一ineomeoftherealestate，theaPPraisalresult15moreobjeetiveandreasonable.3)ComParingwithCostAPProach，thefreshdegreeofthesubjeetProPertywillmanifestthroughtheroomagetarget.几kethetargetoftheroomageasoneofeharaeteristievariables:themodeleandireetlytoobtainitsshadowPrieeandeasilytoobtaintheaccurateaPPraisalresult.4)NotonlytoasuitoftheProPertyaPPraisal，butalsotoalargeamountrealestateaPPraisalthismethod15moreseieniifie，aceurateandquiek.TOthelarge一sealeaPPraisalofthegovernmeni，forexamPle，massivelandsaPPraisal，thismethodalso15oPerational.IV犷Third，eomParingwiththemodemaPPraisalmethods，‘，realestateaPPraisalmethodbasedonself-organizingdatamining”hastwobigsuPeriorities:1)TherequesttosamPledataitself15nothigh.FewerlesdonotaffeettheseientifienatureoftheaPPraisal.2)MelttheelassiealaPPraisaltheoryinioACmodelandenableitoPerational，highereredibleandsuPerior.Keywords:Self-OrganizingDataMiningGMDHCansalityTestingACPattemReeognitionRealEstateAPPraisalV止目录1引言..............……，......................