资源描述:
《[Machine.Learning.in.Action(2012.3)].Peter.Harrington.文字版》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、INACTIONPeterHarringtonMANNINGMachineLearninginActionMachineLearninginActionPETERHARRINGTONMANNINGShelterIslandForonlineinformationandorderingofthisandotherManningbooks,pleasevisitwww.manning.com.Thepublisheroffersdiscountsonthisbookwhenorderedinquantity.Formor
2、einformation,pleasecontactSpecialSalesDepartmentManningPublicationsCo.20BaldwinRoadPOBox261ShelterIsland,NY11964Email:orders@manning.com©2012byManningPublicationsCo.Allrightsreserved.Nopartofthispublicationmaybereproduced,storedinaretrievalsystem,ortransmitted,
3、inanyformorbymeanselectronic,mechanical,photocopying,orotherwise,withoutpriorwrittenpermissionofthepublisher.Manyofthedesignationsusedbymanufacturersandsellerstodistinguishtheirproductsareclaimedastrademarks.Wherethosedesignationsappearinthebook,andManningPubli
4、cationswasawareofatrademarkclaim,thedesignationshavebeenprintedininitialcapsorallcaps.Recognizingtheimportanceofpreservingwhathasbeenwritten,itisManning’spolicytohavethebookswepublishprintedonacid-freepaper,andweexertourbesteffortstothatend.Recognizingalsoourre
5、sponsibilitytoconservetheresourcesofourplanet,Manningbooksareprintedonpaperthatisatleast15percentrecycledandprocessedwithouttheuseofelementalchlorine.ManningPublicationsCo.Developmenteditor:JeffBleiel20BaldwinRoadTechnicalproofreaders:TriciaHoffman,AlexOttPOBox
6、261Copyeditor:LindaRecktenwaldShelterIsland,NY11964Proofreader:MaureenSpencerTypesetter:GordanSalinovicCoverdesigner:MarijaTudorISBN9781617290183PrintedintheUnitedStatesofAmerica12345678910–MAL–171615141312ToJosephandMilobriefcontentsPART1CLASSIFICATION........
7、.......................................................11■Machinelearningbasics32■Classifyingwithk-NearestNeighbors183■Splittingdatasetsonefeatureatatime:decisiontrees374■Classifyingwithprobabilitytheory:naïveBayes615■Logisticregression836■Supportvectormachines
8、1017■ImprovingclassificationwiththeAdaBoostmeta-algorithm129PART2FORECASTINGNUMERICVALUESWITHREGRESSION..............1518■Predictingnumericvalues:regression1539■Tree-basedre