分類算法分析及基于MinerOn Web的設計和實現(xiàn).pdf

1、電子科技大學碩士學位論文分類算法分析及基于MinerOnWeb的設計和實現(xiàn)姓名：張春麗申請學位級別：碩士專業(yè)：計算機軟件與理論指導教師：章毅20090501ABSTRACTABSTRACTInrecentdecades，withthefastdevelopmentofcomputerhardwareandsoftware，especiallythegreatadvanceininteracttechniques，thedatawhich

2、peoplehaveaccumulatedisnowincreasingveryfastThus，datamininghasbecometheresearchhotspotClassification，asonemethodofdatamining，hasacquiredmoreandmoreconcernsbecauseofitsextensiveusageThisthesisfocusonseveralproblemsinclass

3、ificationalgorithmItcomposedofthreeparts：Themethodofclassificationmodelapplications，modelvisualizationandtheimprovedID3Algorithm1)Themethodofclassificationmodelapplications：Thetaskofclassificationistopredicttheclassofnew

4、databyusingthesummarizedrulesfromtrainingdataHoweverinthisprocess，thereexistsomedifficulties，forexample，therulesmaybeintricate，oritishardtomakeuseofthebigrules，although，theserulesareeasytounderstandForaboveproblems，thist

5、hesispresentsaclassificationmodelapplicationsystem，whichisdesignedandimplementedontheMinerOnWebdataminingsystemUnderthissystem，peoplecanusetheruleseasilyeventheycan’tcomprehendtherulesThemodelapplicationsystemcontainsfou

6、rprocesses：modelextraction，modelstorage，modelcomparison，modelapplication2)Modelvisualization：ThevisualizationtechnologycanhelppeopleunderstandmodelstructureandtheresultsofmodelapplicationeasilyItisusedinthemodelapplicati

7、onstageinthispaperFurthermorethemodelvisualizationandmodelapplicationresultsvisualizationaredesignedandimplementedintheMinerOnWebdataminingsystem3)TheimprovedID3algorithm：ID3algorithmisadecisiontreealgorithm，whichisimpor

8、tantinthefieldofmachinelearningTheconceptofinformationgainisproposedbyQuinlaninID3algorithmInformationgainistheselectioncriteriaofthebestsplittingattributeforinducingdecisiontreeNevertheless，thisalgorithmhassomedrawbacks