本文關(guān)鍵詞：基于大數(shù)據(jù)挖掘的手機(jī)用戶(hù)換機(jī)行為研究　出處：《天津商業(yè)大學(xué)》2017年碩士論文　論文類(lèi)型：學(xué)位論文

  更多相關(guān)文章： 換機(jī)行為 抽樣技術(shù) 探索性分析 可視化技術(shù) 數(shù)據(jù)挖掘

【摘要】：隨著互聯(lián)網(wǎng)技術(shù)的飛速發(fā)展和計(jì)算機(jī)應(yīng)用的普及深化,手機(jī)行業(yè)的數(shù)據(jù)采集能力得到了前所未有的提高,獲得并積累了浩如煙海的移動(dòng)通訊大數(shù)據(jù)。通過(guò)大數(shù)據(jù)挖掘技術(shù),研究手機(jī)用戶(hù)的換機(jī)特征及換機(jī)行為的影響因素具有十分重要的意義,它能夠使我們深入地把握用戶(hù)需求,實(shí)現(xiàn)精準(zhǔn)營(yíng)銷(xiāo)�；谶@一背景,本文開(kāi)展了基于大數(shù)據(jù)挖掘的手機(jī)用戶(hù)換機(jī)行為研究。首先,依托某省公司電信運(yùn)營(yíng)商經(jīng)分系統(tǒng),獲得了某段時(shí)期內(nèi)的手機(jī)用戶(hù)換機(jī)數(shù)據(jù),在界定手機(jī)用戶(hù)換機(jī)行為內(nèi)涵、構(gòu)建指標(biāo)體系的基礎(chǔ)上,按照簡(jiǎn)單隨機(jī)抽樣的方法,得到了Kullback-Leibe信息量達(dá)到0.99995以上的樣本,并對(duì)樣本數(shù)據(jù)進(jìn)行了數(shù)據(jù)管理、缺失值處理、離群點(diǎn)檢測(cè)等工作,保證了數(shù)據(jù)質(zhì)量;然后,利用可視化技術(shù)對(duì)換機(jī)前后的手機(jī)品牌特征、手機(jī)性能特征及用戶(hù)特征進(jìn)行了探索性分析;接著,本文分別采用決策樹(shù)、Adaboost、Bagging及隨機(jī)森林算法對(duì)數(shù)據(jù)進(jìn)行訓(xùn)練,建立預(yù)測(cè)模型,得出了各指標(biāo)對(duì)換機(jī)行為的影響程度,并根據(jù)十折交叉驗(yàn)證的平均誤判率對(duì)不同的模型結(jié)果進(jìn)行了分析和比較,其中Adaboost及隨機(jī)森林模型較優(yōu);最后,綜合各模型得出的指標(biāo)重要性排序,結(jié)合實(shí)際,總結(jié)了各指標(biāo)對(duì)換機(jī)行為的影響情況,并提出了手機(jī)廠商要注重品牌建設(shè)、加大技術(shù)投入和創(chuàng)新力度等建議。本文以大量的真實(shí)數(shù)據(jù)為第一手資料,采用定性分析與定量分析相結(jié)合的研究方法,通過(guò)可視化技術(shù)對(duì)手機(jī)用戶(hù)的換機(jī)行為進(jìn)行了深入的研究,在此基礎(chǔ)上,引入了數(shù)據(jù)挖掘中的多種算法分析了換機(jī)行為的影響因素,并提出了一些生產(chǎn)、銷(xiāo)售建議,具有一定的創(chuàng)新性。但是本文在指標(biāo)選取、數(shù)據(jù)抽樣過(guò)程中還存在一些有待改進(jìn)的地方,可在以后的研究工作中加以完善。
[Abstract]:With the rapid development and popularization of computer Internet technology applications deepening, data acquisition ability of the mobile phone industry has been hitherto unknown rise, and accumulated a large data mobile communication. Through the multitude of data mining technology, has very important significance to study the factors affecting the behavior of replacement of mobile phone users characteristics and replacement, it can make we deeply grasp the needs of users, to achieve precision marketing. Based on this background, this paper carried out the research of data mining based on the behavior of mobile phone users switch. First of all, relying on a company of telecom operators to manage the system, obtained within a certain period of replacement Mobile phone users, mobile phone users switch behavior in defining the connotation, construction of basic indicators the system, in accordance with the simple random sampling method, the information content of Kullback-Leibe reached more than 0.99995 of the sample, and the number of samples According to the data management, missing data processing, outlier detection work, to ensure the quality of data; then, mobile phone brand characteristics of replacement before and after the application of visualization technology, an exploratory analysis of mobile phone performance characteristics and user characteristics; then, this paper uses decision tree, Adaboost, Bagging and random forest algorithm for training the data to establish the prediction model, the influence degree of each index on the replacement behavior, and according to the different model results are analyzed and compared with the average error rate of ten fold cross validation, including Adaboost and random forest model is better; finally, the index of importance, the comprehensive model combined with practice, summed up the analysis of the influence of replacement behavior, and put forward the mobile phone manufacturers to focus on brand building, increase technology investment and innovation and so on. Based on a large number of real data for the One hand, adopting the research method of combining qualitative analysis and quantitative analysis, through the switching of visualization technology to the mobile phone users are studied, on this basis, the introduction of a variety of data mining algorithms to analyze the factors affecting the replacement behavior, and puts forward some suggestions of production, sales, innovative sure. But based on the index selection, there are still some room for improvement in the process of sampling data, which could be improved in future studies.

【學(xué)位授予單位】：天津商業(yè)大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類(lèi)號(hào)】：TP311.13;C913.4

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 劉力凱;王國(guó)胤;鄧維斌;;優(yōu)勢(shì)關(guān)系粗糙集的移動(dòng)用戶(hù)換機(jī)預(yù)測(cè)方法[J];小型微型計(jì)算機(jī)系統(tǒng);2015年08期

2 熊冰妍;王國(guó)胤;鄧維斌;;分級(jí)式代價(jià)敏感決策樹(shù)及其在手機(jī)換機(jī)預(yù)測(cè)中的應(yīng)用[J];山東大學(xué)學(xué)報(bào)(工學(xué)版);2015年05期

3 劉彬;焦麗紅;朱東昭;;利用大數(shù)據(jù)洞察客戶(hù)換機(jī)傾向評(píng)估模型研究與分析[J];通信管理與技術(shù);2015年03期

4 李毅;米子川;;大數(shù)據(jù)挖掘的均勻抽樣設(shè)計(jì)及數(shù)值分析[J];統(tǒng)計(jì)與信息論壇;2015年04期

5 鄭英豪;;大數(shù)據(jù)與企業(yè)大數(shù)據(jù)的來(lái)源[J];上海經(jīng)濟(jì);2015年Z1期

6 田紀(jì)軍;黃艷;秦思;廖振松;;基于大數(shù)據(jù)挖掘的終端感知與換機(jī)應(yīng)用研究[J];信息通信;2014年09期

7 魏國(guó)華;康志英;;以客戶(hù)需求為導(dǎo)向的定制終端潛在客戶(hù)挖掘模型研究[J];信息安全與技術(shù);2014年03期

8 朱建平;章貴軍;劉曉葳;;大數(shù)據(jù)時(shí)代下數(shù)據(jù)分析理念的辨析[J];統(tǒng)計(jì)研究;2014年02期

9 趙睿;;基于聚類(lèi)分析法的暢銷(xiāo)手機(jī)分析[J];科技與企業(yè);2014年02期

10 張書(shū)成;;基于“大數(shù)據(jù)”戰(zhàn)略的信息化與新型城鎮(zhèn)化發(fā)展[J];中國(guó)建設(shè)信息;2013年23期

相關(guān)重要報(bào)紙文章 前2條

1 龔燕玲;;智能手機(jī)全球銷(xiāo)量回暖 新興市場(chǎng)掀換機(jī)潮[N];通信信息報(bào);2015年

2 李學(xué)芳;;手機(jī)換機(jī)消費(fèi)時(shí)代到來(lái)[N];中國(guó)經(jīng)營(yíng)報(bào);2008年

相關(guān)碩士學(xué)位論文 前4條

1 魏炎利;自主品牌價(jià)值評(píng)價(jià)的實(shí)證研究[D];廣東工業(yè)大學(xué);2011年

2 闞麗娟;手機(jī)品牌顧客忠誠(chéng)度影響因素實(shí)證研究[D];西安理工大學(xué);2008年

3 張靈明;品牌忠誠(chéng)及其影響因素研究[D];廈門(mén)大學(xué);2006年

4 劉愷;在線民用電表抽樣檢驗(yàn)方法研究與實(shí)證分析[D];重慶大學(xué);2004年

，

本文編號(hào)：1385634