本文關(guān)鍵詞：聚類分析在港口客戶細(xì)分中的應(yīng)用　出處：《北京交通大學(xué)》2015年碩士論文　論文類型：學(xué)位論文

  更多相關(guān)文章： K-means算法 AP算法 PSO算法 港口客戶細(xì)分

【摘要】：隨著國內(nèi)外港口競爭不斷加劇和港口自身業(yè)務(wù)的發(fā)展,要求國內(nèi)港口企業(yè)的運(yùn)營模式,必須逐步向以信息為基礎(chǔ)、以數(shù)據(jù)為中心、以客戶為中心的國際先進(jìn)模式進(jìn)行轉(zhuǎn)變,而實(shí)現(xiàn)這種科學(xué)經(jīng)營模式的前提需要進(jìn)行客戶細(xì)分工作的研究。目前中國港口企業(yè)進(jìn)行客戶細(xì)分的方法還是基于統(tǒng)計(jì)或者基于經(jīng)驗(yàn)的簡單分類方法,并沒有實(shí)現(xiàn)企業(yè)與客戶之間真正的信息交互,無法滿足針對不同客戶需求而提供不同的服務(wù)策略。 聚類分析作為數(shù)據(jù)挖掘技術(shù)中的一種重要方法,已經(jīng)成為該領(lǐng)域中一個非常重要的研究內(nèi)容。聚類分析是在沒有任何先驗(yàn)知識的情況下將一批樣本數(shù)據(jù)(或變量)按照它們在性質(zhì)上的親疏程度自動進(jìn)行分類,最終能夠?qū)崿F(xiàn)樣本空間的盲分類。其次使用數(shù)據(jù)挖掘聚類分析方法進(jìn)行客戶細(xì)分,不但可以處理幾十、甚至上百個變量,從而對客戶進(jìn)行更精準(zhǔn)的描述,客觀地反映客戶分組內(nèi)的特性并綜合反映客戶多方面的特征；而且還有利于營銷人員更加深入細(xì)致地了解客戶特征,便于實(shí)現(xiàn)對客戶行為變化的動態(tài)跟蹤；進(jìn)而實(shí)現(xiàn)對客戶提供差異化服務(wù),提高客戶的滿意度和忠誠度,使企業(yè)創(chuàng)造更多價(jià)值。 本文在現(xiàn)有的港口信息化背景下,首先闡述了在信息化推進(jìn)到現(xiàn)今的階段港口生產(chǎn)數(shù)據(jù)對于分析與挖掘功能的迫切需求和使用數(shù)據(jù)挖掘技術(shù)的必要性。然后對客戶細(xì)分基本理論、聚類分析方法應(yīng)用于客戶細(xì)分的基本理論以及相關(guān)的聚類分析算法做了詳細(xì)的概述,為后文在進(jìn)行客戶細(xì)分中應(yīng)用聚類分析方法奠定了理論基礎(chǔ)。分析港口客戶數(shù)據(jù)庫的情況,選擇和構(gòu)造了港口客戶細(xì)分所需要的屬性,并對其進(jìn)行預(yù)處理,為客戶細(xì)分研究的展開做好數(shù)據(jù)準(zhǔn)備。其次著重分析了傳統(tǒng)的經(jīng)典聚類算法K-means、AP算法和粒子群3種算法在港口客戶細(xì)分中的不足,提出了融合3種算法優(yōu)點(diǎn)的混合型聚類算法,該算法利用AP算法進(jìn)行K值的選取,并充分利用PSO算法的全局搜索能力強(qiáng)與K-means算法局部搜索能力強(qiáng)等特性,通過實(shí)驗(yàn)驗(yàn)證了本文的算法能夠提高聚類的效果和準(zhǔn)確率,加快算法的收斂速度。最后將改進(jìn)的K-means聚類算法應(yīng)用到港口生產(chǎn)業(yè)務(wù)的管理實(shí)踐之中,對客戶細(xì)分結(jié)果進(jìn)行解釋,分析每類細(xì)分市場的特征,結(jié)合港口的實(shí)際情況,針對現(xiàn)有的客戶,給出相應(yīng)的客戶營銷目標(biāo)與策略,并提出了開發(fā)新客戶市場的建議。
[Abstract]:With the development of domestic and international competition intensifies and their business port port, port requirements of domestic enterprises operating mode, turn to the information based, data centric, customer centered international advanced mode transformation, and realize the premise of this scientific management mode of the research needs of customer segmentation work. At present China port customer segmentation is based on the statistical classification method based on experience or simple, and no real information interaction between enterprises and customers, to meet different customer needs and provide different service strategies.
Clustering analysis is an important method of data mining technology, has become a very important research content in the field. Cluster analysis is without any prior knowledge of the case will be a number of sample data (or variables) according to their degree of affinity in the nature of automatic classification, finally can realize blind classification sample space. Secondly using data mining clustering analysis method for customer segmentation, not only can handle dozens, or even hundreds of variables, and thus a more accurate description of the customer, objectively reflect the characteristics of the customer group and reflect various customer characteristics; but also conducive to the marketing personnel more deeply understand customers features, easy to realize dynamic tracking changes in customer behavior; and can provide customers with differentiated services, improve customer satisfaction and loyalty, the creation of enterprises Make more value.
In this paper, the background of existing port information, firstly expounds the necessity of information in advance to the demand and use of data for the current stage of the port production data mining function analysis and mining technology. Then the basic theory of customer segmentation, clustering analysis method was applied to the basic theory of customer segmentation and clustering analysis algorithm is made. Detailed summary, analysis method lays a theoretical foundation of the application of clustering in customer segmentation. In the analysis of port customer database, select and construct attribute port customer segmentation is needed, and carries on the pretreatment for customer segmentation research on data preparation. Then focuses on the analysis of the classical K-means clustering the traditional algorithm, AP algorithm and particle swarm algorithm in 3 port customer segmentation, we propose a hybrid algorithm has the advantages of integration of 3 kinds of Poly Class of algorithms, the algorithm selects the K value by using AP algorithm, and make full use of the global search ability of PSO algorithm and K-means algorithm local search ability and other characteristics, the experimental results indicate that this algorithm can improve the clustering performance and accuracy, accelerate the convergence of the algorithm. Finally, the improved K-means clustering algorithm to the management practice of port production business, for the interpretation of the results of customer segmentation, analysis of each type of market characteristics, combined with the actual situation of the port, for existing customers, the corresponding customer marketing objectives and strategies, and put forward to develop the new customer market.

【學(xué)位授予單位】：北京交通大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2015
【分類號】：F552.6;F274;F224

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 孫娟;張秀梅;;淺談數(shù)據(jù)挖掘理論與技術(shù)[J];辦公自動化;2008年16期

2 齊佳音,李懷祖;客戶關(guān)系管理(CRM)的體系框架分析[J];工業(yè)工程;2002年01期

3 劉悅婷;李嵐;;基于自適應(yīng)權(quán)重的粒子群和K均值混合聚類算法研究[J];甘肅科學(xué)學(xué)報(bào);2010年04期

4 彭志寧;;港口企業(yè)實(shí)施CRM初探[J];中國港口;2006年08期

5 鄧曉懿;金淳;j口良之;韓慶平;;移動商務(wù)中面向客戶細(xì)分的KSP混合聚類算法[J];管理科學(xué);2011年04期

6 胡建秀;曾建潮;;二階微粒群算法[J];計(jì)算機(jī)研究與發(fā)展;2007年11期

7 向堅(jiān)持;陳曉紅;;SaaS模式的中小企業(yè)客戶關(guān)系管理研究[J];計(jì)算機(jī)工程與應(yīng)用;2009年19期

8 傅濤;孫亞民;;基于PSO的k-means算法及其在網(wǎng)絡(luò)入侵檢測中的應(yīng)用[J];計(jì)算機(jī)科學(xué);2011年05期

9 郝先臣,張德干,朱紅艷,趙海;模糊聚類挖掘方法在電子商務(wù)中的應(yīng)用研究[J];計(jì)算機(jī)工程;2002年01期

10 袁方;周志勇;宋鑫;;初始聚類中心優(yōu)化的k-means算法[J];計(jì)算機(jī)工程;2007年03期

，

本文編號：1380547