铁路货运大数据平台下基于聚类的客户细分应用研究

发布时间：2018-06-15 02:25

本文选题：大数据 + 客户细分　；参考：《北京交通大学》2015年硕士论文

【摘要】：近年来,我国铁路货运信息化建设取得了很大的突破和成果,但沉淀的大量货运数据缺乏有效的管理利用,开展大数据技术在铁路货运业务上的数据挖掘研究具有重要的应用价值。客户细分是货运营销的基础,能够更好地识别客户群体,合理地配置企业资源,为企业创造更大的利润。但目前铁路货运的客户细分采用基于经验和统计的简单划分的方法,不能准确区分客户类别,无法有效地支持营销决策。本文将客户细分的常用方法RFM模型做出改进,并与聚类挖掘算法相结合,为铁路货运海量数据下复杂的客户细分问题提供了新的解决方法。本文的主要工作包含以下几个方面： (1)针对铁路货运的特点,对传统的客户细分方法RFM模型做了改进,提出了KFM模型。 (2)由于传统的K-means聚类算法存在对初始聚类中心敏感且容易陷入局部最优的缺点,本文提出了改进的K-means聚类算法。实验表明改进后的算法提高了客户细分的准确率。 (3)将KFM模型与改进后的K-means聚类算法相结合,利用铁路电子商务系统的货运数据进行了客户细分。细分结果很好地展现了各类客户的特征,弥补了传统的基于RFM模型的客户细分对数据挖掘不够深入的缺陷。 (4)在Hadoop大数据平台下,实现了数据标准化方法和K-means聚类算法基于MapReduce的并行化。实验表明基于MapReduce的并行化提升了算法的性能,能胜任大量数据分析处理任务。本文将聚类挖掘技术应用于铁路货运大数据平台下的客户细分,确定不同价值和行为倾向的客户类别,为企业展现出客户所属类别,从而进行针对性管理,有利于货运部门的精准化营销决策。
[Abstract]:In recent years, great breakthroughs and achievements have been made in the construction of railway freight information in China, but a large number of freight data precipitated lack of effective management and utilization. It has important application value to develop data mining research of big data technology in railway freight business. Customer segmentation is the basis of freight marketing, which can better identify customer groups, reasonably allocate enterprise resources, and create greater profits for enterprises. However, the current customer segmentation of railway freight is based on the simple division method based on experience and statistics, which can not accurately distinguish customer categories, and can not effectively support marketing decisions. In this paper, the RFM model of customer segmentation is improved and combined with clustering mining algorithm, which provides a new solution to the complex customer segmentation problem under the massive data of railway freight transport. The main work of this paper includes the following aspects: 1) according to the characteristics of railway freight, the traditional customer segmentation method RFM model is improved. Because the traditional K-means clustering algorithm is sensitive to the initial clustering center and easy to fall into local optimum, this paper proposes an improved K-means clustering algorithm. Experiments show that the improved algorithm improves the accuracy of customer segmentation. (3) the KFM model is combined with the improved K-means clustering algorithm, and the freight data of railway e-commerce system is used to segment customers. The segmentation results show the characteristics of all kinds of customers and make up for the defects of traditional RFM-based customer segmentation which is not deep enough for data mining. Data standardization method and K-means clustering algorithm are implemented based on MapReduce parallelization. Experiments show that the parallelization based on MapReduce can improve the performance of the algorithm and be able to deal with a large number of data analysis tasks. In this paper, clustering mining technology is applied to customer segmentation of railway freight big data platform, and customer categories with different values and behavioral tendencies are determined to show customer categories for enterprises, so as to carry out targeted management. It is beneficial to the precision marketing decision of freight department.
【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】：TP311.13

【参考文献】