數(shù)據(jù)挖掘在電信客戶流失中的應(yīng)用

《數(shù)據(jù)挖掘在電信客戶流失中的應(yīng)用》由會(huì)員上傳分享，免費(fèi)在線閱讀，更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。

1、哈爾濱工業(yè)大學(xué)管理學(xué)碩士學(xué)位論文摘要隨著數(shù)據(jù)挖掘技術(shù)的不斷發(fā)展，人們?cè)絹碓疥P(guān)心它在各個(gè)領(lǐng)域的應(yīng)用，它是利用已有的歷史數(shù)據(jù)通過建立模型的方法找出隱含的業(yè)務(wù)規(guī)則。電信行業(yè)是典型的數(shù)據(jù)密集行業(yè)，而且目前電信運(yùn)營商們都在經(jīng)歷著不同程度的客戶流失，所以運(yùn)用數(shù)據(jù)挖掘技術(shù)找出電信流失客戶的特征顯得尤為重要。本文的研究目的就是應(yīng)用數(shù)據(jù)挖掘技術(shù)發(fā)現(xiàn)流失客戶的特征，建立流失預(yù)測(cè)模型，應(yīng)用模型預(yù)測(cè)流失客戶名單，針對(duì)流失客戶制定有效的挽留策略。本文的主要工作包括以下幾個(gè)方面：首先從業(yè)務(wù)角度，對(duì)電信行業(yè)的業(yè)務(wù)流程和電信客戶流失原因進(jìn)行分析，將客戶流失原因和電信行業(yè)的業(yè)務(wù)系統(tǒng)相接合，總結(jié)出與客戶流失相關(guān)的數(shù)據(jù)

2、資料，完成從商業(yè)角度向數(shù)據(jù)角度的轉(zhuǎn)換。然后從技術(shù)角度，總結(jié)了數(shù)據(jù)挖掘技術(shù)的理論與方法。詳細(xì)描述了神經(jīng)網(wǎng)絡(luò)挖掘算法和決策樹算法的原理及過程，為后面章節(jié)的實(shí)際建模打下鋪墊。接著在前面的基礎(chǔ)上又介紹了電信行業(yè)客戶流失模型建立的原則，探討了數(shù)據(jù)挖掘過程的標(biāo)準(zhǔn)化模型。詳細(xì)描述了CRISP-DM標(biāo)準(zhǔn)數(shù)據(jù)挖掘過程的6個(gè)階段，并結(jié)合電信行業(yè)自身的特點(diǎn)，論述了商業(yè)理解、數(shù)據(jù)理解、數(shù)據(jù)準(zhǔn)備、建立模型、模型評(píng)估、模型應(yīng)用這六個(gè)階段的工作。最后又詳細(xì)探討了流失客戶的挽留價(jià)值，詳細(xì)描述了電信客戶生命周期各個(gè)階段的特點(diǎn)，分析客戶流失的原因，針對(duì)流失客戶制定有效的客戶挽留策略。關(guān)鍵詞數(shù)據(jù)挖掘；客戶流失；跨行業(yè)的

3、標(biāo)準(zhǔn)數(shù)據(jù)挖掘流程；分類回歸樹；挽留價(jià)值-I-哈爾濱工業(yè)大學(xué)管理學(xué)碩士學(xué)位論文AbstractWiththedevelopmentofdataminingtechnology,peoplebecomemoreandmoreinterestedinitsapplicationinvariousfields.Dataminingistofindoutthehiddenbusinessrulesthroughtheestablishmentofthemodelbyusinghistoricaldata.Thetelecomindustryisatypicaldata-intensivein

4、dustry,andcurrentlytelecomoperatorsaresufferingdifferentdegreesofcustomerschurn.Therefore,itisparticularlyimportanttousedataminingtoidentifythecharacteristicsofthechurncustomer.Thepurposeofthispaperistodetectthecharacteristicsofthelostcustomer,establishchurnpredictionmodel,establishthelostcust

5、omerlistbyusingthemodel,andmakesaneffectivedetainmentstrategyagainstthechurncustomers.Themaincontentsofthispaperincludethefollowingaspects:Firstly,fromthebusinessview,thispaperanalysesthebusinessprocessoftelecomindustryandthereasonofcustomerchurn.Thenthispapercombinesthereasonofcustomerchurnan

6、dthetelecombusinesssystem,achievingtheconversionfromcommerceviewtodataview.Secondly,fromthetechnicalview,thispapersummarizesthetheoryandmethodofdatamininganddescribesthealgorithmofneuralnetworkanddecisiontreeindetail.Then,thispaperdescribestheprinciplesofestablishingthemodelofthetelecomindustr

7、ycustomerchurn,anddiscussesthestandarddataminingprocess.ThispaperdescribesthesixstagesofCRISP-DMstandarddataminingprocess,anddiscussesthebusinessunderstanding,dataunderstanding,datapreparation,modelbuilding,modelevaluation,andmodelimple