資源描述:
《基于企業(yè)數(shù)據(jù)倉庫的數(shù)據(jù)挖掘在電信行業(yè)中的研究》由會員上傳分享,免費在線閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、蒸簍饕蘩一業(yè)名師期摘要近年來電信的業(yè)務(wù)是全球經(jīng)濟中增長最快的,同時也是競爭最激烈的。電信如何在眾多企業(yè)中獲得消費者青睞,提高企業(yè)的分析能力,提高企業(yè)的市場競爭力以及維持市場領(lǐng)導(dǎo)地位,是目前最嚴峻的考驗。競爭的加劇使得數(shù)據(jù)倉庫作為決策分析支撐的數(shù)據(jù)平臺日益盛行,并在該平臺上進行數(shù)據(jù)挖掘分析的手段也日漸普遍。本文從闡述數(shù)據(jù)倉庫和數(shù)據(jù)挖掘的概念出發(fā),介紹了企業(yè)數(shù)據(jù)倉庫的構(gòu)建,并提出了電信系統(tǒng)的企業(yè)數(shù)據(jù)倉庫。接著在企業(yè)數(shù)據(jù)倉庫的基礎(chǔ)上提出了適應(yīng)于電信行業(yè)數(shù)據(jù)挖掘模塊的設(shè)計實現(xiàn),將適合電信行業(yè)的數(shù)據(jù)挖掘模塊和企業(yè)數(shù)據(jù)倉庫結(jié)合起來,滿足電信行業(yè)的數(shù)據(jù)挖掘需求。在對電信行業(yè)企業(yè)數(shù)據(jù)倉庫
2、進行了深入的研究之后,針對電信行業(yè)中常見的客戶呼叫模式的關(guān)聯(lián)分析和電信大客戶特征的聚類識別,本文提出了基于分區(qū)的散列算法(HashPartitionAlgorithm)和基于K-means算法的遺傳算法?;诜謪^(qū)的散列算法(HashPartitionAlgorithm)充分考慮了電信行業(yè)中基于海量數(shù)據(jù)中的數(shù)據(jù)挖掘性能,實現(xiàn)了對數(shù)據(jù)分區(qū)的設(shè)計,極大地減少了數(shù)據(jù)庫的掃描次數(shù),同時很好地實現(xiàn)了散列技術(shù)和分區(qū)技術(shù)的融合,并給出了算法的具體實現(xiàn)過程和基于企業(yè)數(shù)據(jù)倉庫的實驗結(jié)果?;贙-means算法的遺傳算法實現(xiàn)了適合海量數(shù)據(jù)挖掘的遺傳算法和K-means算法的混合。該算法通過遺傳
3、算法,可以顯著地降低對數(shù)據(jù)庫的瀏覽次數(shù),提高算法性能,并能夠準確充分地反映大客戶的特征,從而實現(xiàn)對大客戶特征的聚類。關(guān)鍵字:數(shù)據(jù)倉庫數(shù)據(jù)挖掘關(guān)聯(lián)分析散列算法K-means算法聚類遺傳算法TheResearchofDataMiningBasedonEnterpriseData紓白rehouseofTelecomABSTRACTbusinessoftelecomisthefastest-risingintheglobaleconomyinrecentyearsand“isalsothemosthotlycompetitiveatthesametime.Howtelecomge
4、tsconsumerstofavorinnumerousenterprisestoimprovetheanalyticalcapacityandthemarketcompetitivenessofenterprisesandmaintainstheleadingpositionofmarketisthemostseveretestatpresent.a(chǎn)ggravationofthecompetitionmakesthedatawarehouseanalysisasthedataplatformsupportedtomakepolicyprevailsdaybyday,an
5、ddataminingattheplatformisinereasin醇yused.Firstly,inthisthesisweexplaintheconceptofdatawarehouseanddataminingandintroducetheconstructionoftheenterprisedatawarehouse.Thenweproposetheenterprisedatawarehouseofthetelecomsystem.Secondly,weproposethemoduledesignandrealizationofdataminingbased01
6、1enterprisedatawarehousewhichcombinethedataminingmoduleandenterprisedatawarehousetomeetthedemandofdataminingfortelecom.WeproposeHashPartitionAlgorithmandGeneticAlgorithmbasedonK-meansaccordingtoaSsociationanalysisofcallmodeofcustomersandclus蜘ngofVIPaftercarryingonde印researchtoenterpriseda
7、tumwarehouse.HashPartitionAlgorithmtakesperformanceofdataminingintoconsiderationonthebasisofmassdataofTelecom.Ithasrealizedreducingthenumberoftimesofscanningdatabasegreatlybypartitiondesignofdata.Atthesanletime,havingrealizedtheintegrationofhashalgorithmandpartition