聚类分析在港口客户细分中的应用

发布时间：2018-01-04 23:18

本文关键词：聚类分析在港口客户细分中的应用　出处：《北京交通大学》2015年硕士论文　论文类型：学位论文

【摘要】：随着国内外港口竞争不断加剧和港口自身业务的发展,要求国内港口企业的运营模式,必须逐步向以信息为基础、以数据为中心、以客户为中心的国际先进模式进行转变,而实现这种科学经营模式的前提需要进行客户细分工作的研究。目前中国港口企业进行客户细分的方法还是基于统计或者基于经验的简单分类方法,并没有实现企业与客户之间真正的信息交互,无法满足针对不同客户需求而提供不同的服务策略。聚类分析作为数据挖掘技术中的一种重要方法,已经成为该领域中一个非常重要的研究内容。聚类分析是在没有任何先验知识的情况下将一批样本数据(或变量)按照它们在性质上的亲疏程度自动进行分类,最终能够实现样本空间的盲分类。其次使用数据挖掘聚类分析方法进行客户细分,不但可以处理几十、甚至上百个变量,从而对客户进行更精准的描述,客观地反映客户分组内的特性并综合反映客户多方面的特征；而且还有利于营销人员更加深入细致地了解客户特征,便于实现对客户行为变化的动态跟踪；进而实现对客户提供差异化服务,提高客户的满意度和忠诚度,使企业创造更多价值。本文在现有的港口信息化背景下,首先阐述了在信息化推进到现今的阶段港口生产数据对于分析与挖掘功能的迫切需求和使用数据挖掘技术的必要性。然后对客户细分基本理论、聚类分析方法应用于客户细分的基本理论以及相关的聚类分析算法做了详细的概述,为后文在进行客户细分中应用聚类分析方法奠定了理论基础。分析港口客户数据库的情况,选择和构造了港口客户细分所需要的属性,并对其进行预处理,为客户细分研究的展开做好数据准备。其次着重分析了传统的经典聚类算法K-means、AP算法和粒子群3种算法在港口客户细分中的不足,提出了融合3种算法优点的混合型聚类算法,该算法利用AP算法进行K值的选取,并充分利用PSO算法的全局搜索能力强与K-means算法局部搜索能力强等特性,通过实验验证了本文的算法能够提高聚类的效果和准确率,加快算法的收敛速度。最后将改进的K-means聚类算法应用到港口生产业务的管理实践之中,对客户细分结果进行解释,分析每类细分市场的特征,结合港口的实际情况,针对现有的客户,给出相应的客户营销目标与策略,并提出了开发新客户市场的建议。
[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.

【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F552.6;F274;F224

【参考文献】