决策树分类算法及其应用

《决策树分类算法及其应用》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、Y1404797分类号：UDC：密级：学校代号：10150学号：20052205峡童交通戈乎硕士学位论文决策树分类算法及其应用decisiontreealgorithmandapplication学生姓名：导师及职称：学科门类：专业名称：程晓兰宋旭东副教授工学计算机应用技术研究方向：数据挖掘与决策支持系统申请学位级别：论文答辩日期：学位授予单位：硕士2007年11月17日大连交通大学大连交通大学T学硕十学位论文AbstractWiththeconstructionofenterpriseinfoIrmation，dataw

2、arehouseanddecisionsupportsystemtechnologyintheenterprisehasundergoneanunprecedentedapplication．Howwillthedecisionsupportsystemapplicationofdatamimngmethodstoenterprisesbecomethefocusofthestudy?Themainthesisisonthekeytechnologiesofthedecisiontreealgorithmoftheclas

3、sificationalgorithmofdatamining．Thisarticlefirstdoesareviewondecisiontreeclassificationalgorithm．Thebriefsummaryisaboutthemaincharacteristic，advantagesanddisadvantages，scopeofapplication，presentimprovementstate，applicationandexpectingoftypicalcategorizedalgorithms

4、ofdecisiontree．Alongwiththedataprocessingtechnology’Srapiddevelopment，thedatascalewhichneedstobeprocessedhasbeenmuchbiggerthanbefore，italreadyhasdevelopedfromtheinitialsmalldatabasetothepresentlarge-scaledatabaseandthedatawarehouse．Thenvalidity，accuracyandspaceoft

5、hedataminingbecameamajorconsiderationinthecharacteristics．Aftertheformerfaceofthetypicaldecisiontreeclassificationalgorithmresearch，thesamplingtechnologywillbeintroducedintotheDecisionTreeAlgorithmC4．5，makingsmalldatasetseffectivealgorithmcanalsogivenacertaincorre

6、ctrulesforthelargedatasets．ChoosestandarddatabaseofUCImachinelearninglibraryasdatasource，usedecisiontreeC4．5algorithmbasedonsamplingtodigoutclassificationrules．Testingshowsthatthemethodcansignificantlyimprovetheefficiencyofdataminingonconditionthatobtainssatisfact

7、orycorrectness．Immediatelycombineoneconcreteapplicationbackgroundontheironandsteelenterprises，theimprovedalgorithmwillbeusedintwomajoraspects：ironandsteelenterpriseskeyprocessesofproductioncostsandalossofironandsteelenterprises．Thefirstapplicationregardscraftroute

8、asthebreakthroughpoint，combinethecostanalysisprojectsofenterprises，dodatawarehousemodelingforproductioncostprocesses．AdoptimproveddecisiontreealgorithmC