孤立点检测在移动通信数据分析上的分析与应用

《孤立点检测在移动通信数据分析上的分析与应用》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、广东工业大学硕士学位论文析和孤立点检测技术，应用剪枝方法和模糊处理等技术，将孤立点检测方法应用到移动无线网络通信数据分析。针对方法的检测准确率和检测时间两个性能进行实验，实验结果表明该方法具有很好的检测效果，整体-¨ILT．厶匕I=J匕比较高。关键字：划分；孤立点检测；无线网络；剪枝；模糊处理；平衡隶属度ABSTRACTPotentiallymeaningfulknowledgecanfoundfrommassivedatabyDataminingtechniques，it'sanimportantchallengetofindfe

2、wabnormalbehavioroftheobjectfromanumberofdataandfindameaningfulpatternfromthesebehaviors．ThereareoftensomedataPointscalledoutliers，whicharedifferentfromnormalPointsorthePatternofdatasetinthereallife．AsweknowthatthenormalbehavioriSmuchmorethanabnormalbehavior，butasmalln

3、umberofabnormalbehaviorsmayhavehiddenaninterestingknowledge．Therefore，studyofoutlierhassomebasistheoreticalandpracticalsignificance．InthispapeLmakesadetailstudyforoutlierdetectionmethod，combinedwithclusteranalysisandoutlierdetectiontechnologyindatamining，atthesametime，

4、makesananalysisforresearchandapplicationoftheoutlierdetectionmethodathomeandabroad，givesthemodelofoutlierdetectionapplication．Thenintroducetheuseofdatapreprocessing，itcanimprovetheoverallperformanceofdetection，andfinallygivespartition-basedoutlierdetectionmethod，applie

5、dtothefieldofwirelessnetworkcommunicationdataanalysis．Themainresearchworkincludes：1．Makingananalysisfortheresearchbackgroundandresearchstatusofoutlierdetectionmethodathomeandabroad，itsrangeofapplicationsaswellasitsadvantagesanddisadvantagesareanalyzedindetail，gettherel

6、evantconclusions，outlierdetectiontechnologyisappliedtothemobilecommunicationfieldofdataanalysis．2．Accordingtotheknowledgeandtheapplicationofoutlierdetection，giventhebasicmodelofoutlierdetection，andadetailedanalysisofeachofitscomponentparts．3．Becauseoftheinconsistentspe

7、cificationoftheoriginaldatasetandlargedata，makingadetailedanalysisoftheoriginalcommunicationdatasets，andthroughadetaileddata-preprocessingofthedatasetdatacleaning，attributefieldtoselect，广东工业大学硕士学位论文formatconversionandfinallyobtainhighqualitytobedetecteddataset，inthiswa

8、ycanimprovetheefficiencyoftheoutlierdetection．4．Accordingtothemodelofoutlierdetection，wecombinedwiththeneedsoftheoutl