混合屬性數據集異常檢測新方法

1、中國石油大學（華東）碩士學位論文混合屬性數據集異常檢測新方法姓名：劉緒棟申請學位級別：碩士專業(yè)：計算機科學與技術指導教師：龔安201205LiuXudong(ComputerScience&Technology)DirectedbyAssociateProfGongAnAbstractInthefeldofnaturehumansocietyanddataset，therearemanyunusualtimesoroWects，thes

2、eBnusualeventsorobjectshaveextraordinaryimportance，itmayleadtothediscoveryofnewinformationandknowledgeValuableinformationandknowledgedonotgetthefulldevelopmentandutilization，whichurgentlyrequirepeopletotakeeffectivemeasu

3、restominingtheexceptionimplicitknowledge，inordertoobtainusefulinformationandknowledge；outlierminingtechnologycameintobeingWiththecontinuousdevelopmentofanomalydetectionalotofdifferentareasofapplicationalgorithmsisdevelop

4、edInthesealgorithms，therearesomeproblemsaffectingthealgorithmaccuratelyrateandefficiencySuchasparameterautomationdealingwithmixed—attributedatasetdiversitymeasureandthresholdsettingsandSOonThetwoaspectsofdiversitymeasure

5、andtheabnormalfactorareimproved；allanomalydetectionmethodbasedonthelocaldensityofthemixedattributeisputforwardForamixed—attributediversitymeasure，numericalpropertiesandclassificationofpropertyisdealt麗m血eirdifferences，the

6、finalcombinationThediversitymeasureofthevaluepropertyusingtheEuclideandistanceformulaclassificationofpropertydiversitymeasureweightedHammingdistanceformula；theweightisdefinedasthenumberofclassificationofpropertyisnotequa

7、lbetweeneachobjectattributevaluesThelasttwodifferencesmeasureofthesuln，andtheresultdividedbythenumberofattributes，ameasureoftheaverageofthedifferenceonthefinalproperties，asthedifferencesbetweenthehandlemixedattributesdat

8、asetobjecttomeasureDefinitionofabnormalfactorsusedtocalculatethemixedattributedatafocusonobjectsbetweenlocaldensityandneighborhooddensityfactorthelastoftheabnormalfactorisobtainedfromtheproductofboththedistinctiontheexce