【作者】 肖娟；

【导师】 叶枫； 汪少华；

【作者基本信息】 浙江工业大学 ， 技术经济及管理， 2004， 硕士

【摘要】 数据挖掘指从大量的数据中发现潜在的，有用的知识的过程，是实现现代企业客户关系管理的重要手段之一。数据分类方法作为数据挖掘的一个重要方面，在统计、机器学习、神经网络、专家系统等领域中有着广泛的应用，如医疗诊断、天气预测、信用证实、顾客区分、欺诈甄别等。 现有数据分类方法有统计方法、决策树分类方法、神经网络方法、粗集法等。其中概念层次树作为一种统计类的数据分类方法在面向属性的归纳分类方法中起着重要作用。通过它对数据库中的数据进行分类可以帮助我们发现数据的特征，以更加容易理解的方式总结数据，并且依据面向概念的结构来组织数据。 本文针对已有数据分类算法的不足，采用“变间隔分割初始区间”和“概念层次构建与概念提升同步”的方式“自底向上”地构建概念层次树，提出并实现了基于概念层次树的数据挖掘改进算法。 改进算法针对数值型数据呈不均匀分布的情况，相比原算法有更好的效果。本论文对两种算法在实际数据库上进行测试，给出了数据分析结果，并给出了该改进算法在外贸企业客户关系管理中的应用实例。更多还原

【Abstract】 Data mining which means a kind of process that reveals potential useful knowledge from massive, is one important technology in CRM of modem enterprises. Data classification, as an important aspect of research of DM, now has been successfully applied to medical diagnosis, weather prediction, credit approval, customer segmentation, fraud detection and so on.Many different techniques have been proposed for classification, including statistical approaches, neural networks, decision tree algorithm and rough sets. Conception Hierarchy Tree classifiers which is a statistical approach have played an important role in Attribute-Oriented Induction. It can help us discover the characteristics of data, make them more understandable and organized in concept-oriented structure.This dissertation, in the light of the limitations of existed methods, suggest an algorithm based on Conception Hierarchy Tree for data mining, constructing tree from bottom to top through the method of variedly dividing interval and realizing conception hierarchy construction and conception exaltation isochronously.Improved algorithm has showed great effects compared to original algorithm especially to numerical attributes with asymmetry distribution in database. This paper displays the result of two algorithms testing on actual data and offer an example in CRM of a foreign trade enterprise.更多还原