多元统计方法在市场细分中的应用.pdfVIP

Application of Multivariate Statistical Methods in the Market Segmentation A Dissertation Submitted for the Degree of Master Candidate：Xuhui Supervisor：Prof. Manfa Liang South China University of Technology Guangzhou, China 摘 要 市场细分是制定营销策略的关键依据，就目前而言，我国现行的市场研究中较少涉 及到多维指标的市场细分理论与方法研究，鉴于此，开展“多元统计分析方法在市场细 分中的应用”课题研究有其重要的现实意义。 本论文首先依据通常市场营销学中的市场细分理论，选择调查指标和分析指标，但 由于指标众多而又相互关联，不能清晰细分市场。由此我们通过对原指标进行非典型相 关分析，从中找出一组相互独立且信息贡献最大的指标，以实现模型降维的目的，从而 保证指标对市场划分达到整体协同效应好。接下来本文用新指标对样本聚类，通过对不 同的聚类方法效果比较，获得较佳市场细分效果。利用实际调研的案例——对东莞消费 群体的市场细分研究，本文将人群细分为五类。最后，将分类结果进行统计描述性分析， 揭示细分市场特征和市场营销意义，并用 MDS 技术图示市场细分结果及营销策略依据。 本论文结论得到项目委托企业的认同。 关键字：市场细分；非典型相关分析；聚类分析 I Abstract Market segmentation is the key to formulate marketing strategy ，now in China ，there are less market segmentation theories and methods that related to multi-dimensional indicators in market research, thus, it has important practical value to carry out the research which is Application of Multivariate Statistical Methods in the Market Segmentation. In this paper, first of all, we choose survey indicators and analysis indicators according to the market segmentation theory, but actually there are many indicators and they relate to each other, so market segmentation could not be described clearly. Thus we use Nonlinear Canonical Correlation Analysis to find a set of mutually independent and the largest contribution to the information index, in order to make implementation model dimension reduced, guarantee good index to dividing the overall market synergies. With new index of sampl