搜索
1密级：中图分类号：硕士学位论文论文题目：基于变权理论的群组评价方法研究作者姓名：卞起鹏学科专业：应用统计学研究方向：经济统计指导教师：陈骥提交日期：2017年10月I基于变权理论的群组评价方法研究摘要群组评价的应用已经深入到经济社会生活的各个方面，作为统计综合评价的一个重要分支，群组评价越来越引起社会的关注，已成为各类评比、择优、决策等活动中不可缺少的评价工具。在被广泛应用的同时，群组评价的新思路、新方法日渐增多，但在面对实际评价问题时也会存在诸多问题。权重作为群组评价问题的主要要素之一，是以某种数量形式对比、权衡被评价事物总体中诸因素相对重要程度的量值，很显然，当评价指标及被评价对象都确定时，群组评价的结果就要依赖于权重系数。合理确定权重对决策或评价有着重要意义，不同的权重系数，会导致截然不同的甚至相反的评价结论。因此，权数确定问题是群组评价中十分棘手的问题。在面对一些复杂评价问题时，往往会因为使用“常权”这种赋权方式，得到并不合理的评价结果。另外，在群组评价过程中，由于各评价者的知识水平、个人偏好等有较大差异，他们在作出评价时往往会带有很强的个人偏见，对自己厌恶的方案给予较低的评价；另一方面，因为某些原因，少数评价者会对部分方案给出明显错误的评价，这种情况下使用线性加权法对群租评价信息进行集结，这些不公平或错误的评价往往会导致不合理甚至错误的群组评价结果。因此，本文考虑“变权”的方式进行赋权。笔者这里按时间维的概念，将群组评价分为静态群组评价和动态群组评价，并利用变权的思想，对评价过程中涉及的权重进行确定。对于静态群组评价问题，首先，提出变权效用函数评价方法，并利用二次加权的方法对信息进行集结。然后，考虑变权的优劣效应，提出基于变权理论的自适应赋权机制，通过对群组评价共识度的设计，测算变权后评价共识度是否有提高，是否达到群组一致的目的，进而反应变权的优劣效用。II对于动态群组评价问题，笔者将动态群组评价分为无反馈的多阶段动态群组评价和有反馈的交互式动态群组评价，并分别对这两类典型的动态群组评价问题进行了探讨。针对无反馈的多阶段动态群组评价问题，首先测算评价个体多个阶段的满意偏好，它反映了各专家的真实偏好，然后从考虑各专家的各阶段评价信息的稳定性，对于专家权重，提出基于个体波动效应的变权机制，进而得到最终的群组满意偏好；对于有反馈的交互式动态群组评价问题，针对一般的交互式群组评价中存在的不足，对一般的交互式群组评价方法进行改进，从“横向”考虑各阶段内专家评价信息的关联情况，从“纵向”考虑各阶段间评价个体的交互影响力，对各交互阶段各评价个体进行权重修正，然后考虑各轮次群组评价的离散程度得到各轮次权重，最后综合得到方案的排序。关键词：群组评价；变权；动态群组评价；权重IIIResearchongroupevaluationmethodbasedonVariableWeightTheoryABSTRACTApplicationofcomprehensiveevaluationhasgonedeepintoallaspectsofeconomicandsociallife,asanimportantbranchofstatisticalcomprehensiveevaluation,comprehensiveevaluationismoreandmoreattractedtheattentionofthesociety,hasbecomeindispensabletoallkindsofcompetitions,merit,decision-makingandotheractivitiesintheevaluationtool.Whilewidelyused,thenewideasandmethodsofcomprehensiveevaluationareincreasingdaybyday,buttherearealsomanyproblemsinthefaceofpracticalevaluationproblems.Asoneofthemainfactorsofweightcomprehensiveevaluationproblem,istoweighanumberoffactors,thecomparativeevaluationofthingsingeneraltherelativeimportanceofvalue,obviously,whentheevaluationindexandtheevaluationobjectsaredetermined,thecomprehensiveevaluationresultswilldependontheweightcoefficient.Therationaldeterminationofweightsisimportanttodecisionmakingorevaluation,anddifferentweightswillleadtodifferentorevenoppositeevaluationconclusions.Therefore,theproblemofdeterminingweightsisaverydifficultproblemincomprehensiveevaluation.Inthefaceofsomecomplexevaluationproblems,oftenbecauseoftheuseofconstantweightthiswayofempowerment,andgetunreasonableevaluationresults.Inaddition,thegroupevaluationprocess,duetotheevaluationofthelevelofknowledge,personalpreferencesaredifferent,theyareoftenevaluatedwithstrongpersonalbias,giventhelowerevaluationontheiraversiontothescheme;ontheotherIVhand,forsomereason,afewreviewersofpartoftheprogramgiventheobviouserrorintheevaluation,toaggregatethegroupevaluationinformationbyusinglinearweightedmethod,inthiscase,theseunfairorwrongevaluationoftenleadstounreasonableorevenwronggroupevaluationresults.Therefore,thispaperconsidersthevariableweightapproachtoempowerment.Theauthordividesthegroupevaluationintostaticgroupevaluationanddynamicgroupevaluationaccordingtotheconceptoftimedimension,andusestheideaofvariableweighttodeterminetheweightsintheprocessofevaluation.Forthestaticgroupevaluationproblem,firstofall,thevariableweightutilityfunctionevaluationmethodisproposed,andthetwoweightedmethodisusedtocollecttheinformation.Then,consideringthequalityeffectofvariableweight,putforwardtheadaptivevariableweightingmechanismbasedonthetheoryoftheright,throughthedesignofevaluationofthedegreeofconsensus,measurethevariableweightevaluationconsensusdegreeisimproved,whetherthepurposeofthegroupconsistent,andthenthereactionqualityvariableweightutility.Forthedynamicgroupevaluationproblem,thedynamicgroupevaluationisdividedintomultistagedynamicgroupevaluationwithoutfeedbackandinteractivedynamicgroupevaluationandfeedback,andeachofthetwotypesofdynamicgroupevaluationoftypicalproblemsarediscussed.Aimingattheproblemofmultistagedynamicgroupevaluationfeedback,firstcalculatesthesatisfactionofmultiplestagesoftheevaluationofindividualpreference,itreflectsthetruepreferencesofexperts,andconsideringthestabilityofeachstageoftheexpertsevaluationinformation,theexpertweight,putforwardtherightmechanismbasedonindividualwaveeffect,andthengetthefinalgroupsatisfactionforinteractivedynamicpreference;groupevaluationfeedbackproblems,aimingattheproblemsofinteractivegroupevaluationingeneral,theinteractivegroupevaluationisimproved,consideringthecorrelationevaluationinformationineachstageofexpertsfromthehorizontal,consideringtheinteractionbetweentheindividualevaluationinfluenceeachstagefromvertical,theVevaluationoftheinteractionstageofindividualweightcorrection,andthenconsidertheconsistencyofeachstageofgroupevaluationobtainedineachstageoftheweight.Therankingofthesolutionsisintegrated.KEYWORDS:GroupEvaluation;VariableWeight;DynamicGroupEvaluation;WeightVI目录摘要..............................................................................................................................IABSTRACT.............................................................................................................