资源描述：
《making data analysis expertisebroadly accessible through workflowsby hau》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、MakingDataAnalysisExpertiseBroadlyAccessiblethroughWorkflowsMatheusHauderYolandaGilandRickySethiYanLiuandHyunjoonJoInformationSciencesInstituteInformationSciencesInstituteComputerScienceDepartmentUniversityofSouthernCaliforniaUniversityofSouthernCaliforniaUniversityofSouthernCalifornia4676A
2、dmiraltyWay4676AdmiraltyWay941BloomWalkMarinadelRey,CA90292MarinadelRey,CA90292LosAngeles,CA90089-0781+1-310-822-1511+1-310-822-1511+1-213-740-4371matheus@isi.edugil@isi.edu,rsethi@isi.eduyanliu.cs@usc.edu,hyunjoon@usc.eduABSTRACTcyberobservatories,tocorrelatingon-lineuserbehaviors,toThedem
3、andforadvancedskillsindataanalysisspansmanyareasaggregatingmedicalrecords.Althoughfoundationalknowledgeisofscience,computing,andbusinessanalytics.Thispapertaughtinmajoruniversitiesandcolleges,advanceddataanalyticsdiscusseshownon-expertusersreuseworkflowscreatedbycanonlybeacquiredthroughhand
4、s-onpracticaltraining.Onlyexpertsandrepresentingcomplexdataminingprocessesfortextexposuretoreal-worlddatasetsallowsstudentstolearntheimportanceofpreparingandcleansingthedata,designinganalytics.Theyincludeworkflowsfordocumentclassification,appropriatefeatures,andformulatingthedataminingtasks
5、othatdocumentclustering,andtopicdetection,allassembledfromthedatarevealsphenomenaofinterest.However,theeffortcomponentsavailableinwell-knowntextanalyticssoftwarelibraries.Theworkflowsexposetonon-expertsexpert-levelrequiredtoimplementsuchcomplexmulti-stepdataanalysisknowledgeonhowtheseindivi
6、dualcomponentsneedtobesystemsandexperimentwiththetradeoffsofdifferentalgorithmscombinedwithdatapreparationandfeatureselectionstepstoandfeaturechoicesisdaunting.Formostpracticaldomains,itmaketheunderlyingstatisticallearningalgorithmsmosteffective.cantakeweekstomonthsforastudenttosetupthebasi
7、cTheframeworkallowsnon-expertstoeasilyexperimentwithinfrastructure,andonlythosewhohaveaccesstoexpertstopointdifferentcombinationsofdataanalysisprocesses,representedasthemtotherighthigh-leveldesignchoiceswillendeavoronthistypeoflearning.Asaresult,acquirin
