资源描述：
《the development of a data mining study》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、2ThedevelopmentofadataminingstudyBeforeexaminingthekeypointsofadataminingstudyorprojectinthefollowingchapters,Ishallbrieflyrunthroughthedifferentphasesinthischapter.Eachphaseismentionedhere,eventhoughsomeofthemareoptional(thosecoveredinSections2.5and2.7),andmostofth
2、emcanbehandedovertoaspecialistfirm(exceptforthosecoveredinSection2.1andpartsofSections2.2and2.10).Iwillprovidefurtherdetailsandillustrationsofsomeofthesestepsinthechapteronscoring.Weshallassumethatstatisticalordataminingsoftwarehasalreadybeenacquired;thechapteronsof
3、twaredealswiththecriteriaforchoiceandcomparison.Asageneralrule,thephasesofadataminingprojectareasfollows:.definingtheaims;.listingtheexistingdata;.collectingthedata;.exploringandpreparingthedata;.populationsegmentation;.drawingupandvalidatingthepredictivemodels;.dep
4、loyingthemodels;.trainingthemodelusers;.monitoringthemodels;.enrichingthemodels.DataMiningandStatisticsforDecisionMaking,FirstEdition.Ste´phaneTuffe´ry.Ó2011JohnWiley&Sons,Ltd.Published2011byJohnWiley&Sons,Ltd.26THEDEVELOPMENTOFADATAMININGSTUDY2.1DefiningtheaimsWemu
5、ststartbychoosingthesubject,definingthetargetpopulation(e.g.prospectsandcustomers,customersonly,loyalcustomersonly,allpatients,onlythosepatientswhocanbecuredbythetreatmentundertest),definingthestatisticalentitytobestudied(e.g.aperson,ahouseholdconsistingofspousesonly
6、,ahouseholdincludingdependentchildren,abusinesswithorwithoutitssubsidiaries),definingsomeessentialcriteriaandespeciallythephenom-enontobepredicted,planningtheproject,decidingontheexpectedoperationaluseoftheinformationextractedandthemodelsproduced,andspecifyingtheexp
7、ectedresults.Todothis,wemustarrangeformeetingstobeheldbetweentheclients(riskmanagementandmarketingdepartments,specialists,futureusers,etc.,asappropriate)andtheserviceproviders(statisticiansandITspecialists).Assomedataminingprojectsaremainlyhorizontal,operatingacros
8、sseveraldepartments,itwillbeusefulforthegeneralmanagementtoberepresentedatthisstage,sothattheinevitablearbitrationcantakeplace.Thetopmanagementwi
