资源描述：
《is it just me or does this data smell funny》由会员上传分享，免费在线阅读，更多相关内容在工程资料-天天文库。
1、CHAPTER2IsItJustMe,orDoesThisDataSmellFunny?KevinFinkYouaregivenadatasetofunknownprovenance.Howdoyouknowifthedataisanygood?Itisnotuncommontobehandedadatasetwithoutalotofinformationastowhereitcamefrom,howitwascollected,whatthefieldsmean,andsoon.Infact,it’sprobablymorecommontoreceivedatainthi
2、swaythannot.Inmanycases,thedatahasgonethroughmanyhandsandmultipletransformationssinceitwasgathered,andnobodyreallyknowswhatitallmeansanymore.Inthischapter,I’llwalkyouthroughastep-by-stepapproachtounderstanding,validating,andultimatelyturningadatasetintousableinformation.Inparticular,I’lltal
3、kaboutspecificwaystolookatthedata,andshowsomeexamplesofwhatIlearnedfromdoingso.Asabitofbackground,Ihavebeendealingwithquiteavarietyofdataforthepast25yearsorso.I’vewrittencodetoprocessaccelerometerandhydrophonesignalsforanal‐ysisofdamsandotherlargestructures(asanundergraduatestudentinEnginee
4、ringatHarveyMuddCollege),analyzedrecordingsofcallsfromvariousspeciesofbats(asagraduatestudentinElectricalEngineeringattheUniversityofWashington),builtsys‐temstovisualizeimagingsonardata(asaGraduateResearchAssistantattheAppliedPhysicsLab),usedlargeamountsofcrawledwebcontenttobuildcontentfilt
5、eringsys‐tems(astheco-founderandCTOofN2H2,Inc.),designedintranetsearchsystemsforportalsoftware(atDataChannel),andcombinedmultiplesetsofdirectoryassistancedataintoasearchablewebsite(asCTOatWhitePages.com).Forthepastfiveyearsorso,I’vespentmostofmytimeatDemandMediausingawidevarietyofdatasource
6、stobuildoptimizationsystemsforadvertisingandcontentrecommendationsystems,withvarioussideexcursionsintolarge-scaledata-drivensearchengineoptimization(SEO)andsearchenginemarketing(SEM).5MostofmyexampleswillberelatedtoworkI’vedoneinAdOptimization,ContentRecommendation,SEO,andSEM.Theseareas,asw
7、ithmost,havetheirownterminol‐ogy,soafewtermdefinitionsmaybehelpful.Table2-1.TermDefinitionsTermDefinitionPPCPayPerClick—Internetadvertisingmodelusedtodrivetraffictowebsiteswithapaymentmodelbasedonclicksonadvertisements.Inthedataworld,itisusedmorespecific
