【作者】 尹华；

【导师】 董红斌；

【作者基本信息】 武汉大学 ， 软件工程， 2004， 硕士

【摘要】 当今社会是一个信息爆炸的时代，如何充分利用各种各样的信息为人类服务已显得越来越重要。作为一门新兴的边缘学科——数据挖掘受到了国内外的普遍关注，成为信息系统和计算机科学领域研究最活跃的前沿领域。 本文以广东省重点科技攻关项目“数据仓库开发工具与智能分析平台研究”(A1020103)为背景，在深入调研与分析国内外数据挖掘相关理论与应用和技术文献的基础上，归纳总结了该领域的主要研究内容和关键技术，评述了数据挖掘系统开发相关理论与技术的研究现状、存在问题与发展趋势。利用中国银行的数据仓库(广东华际友天信息科技有限公司提供)作为实验数据，主要研究了数据挖掘过程中的以下几个问题：数据挖掘项目开发过程模型、数据预处理中的数据质量分析标准、数值属性的离散化算法和决策树算法的改进。本论文的主要研究工作与成果有： 1、研究与分析了数据挖掘系统开发所涉及的数据挖掘技术，数据挖掘方法论以及数据挖掘系统等相关理论的研究与应用现状及发展趋势。 2、针对目前的数据挖掘过程模型在实际应用中仅考虑数据挖掘技术，忽略了项目的规划、部署与数据挖掘技术协调的问题，在已有的数据挖掘过程模型基础上提出了基于项目开发的数据挖掘过程模型，该模型将数据挖掘项目开发过程分为项目规划、挖掘准备、挖掘和项目评估四个阶段。详细地描述了针对数据挖掘项目开发各阶段(尤其是项目规划阶段)的具体任务，使数据挖掘技术能够更好地应用解决商业问题。 3、完成基于数据挖掘分类技术的信用卡挖掘系统的研究与构建，验证了基于项目的数据挖掘过程理论。针对银行数据中所存在的质量问题提供数据质量分析标准规范；针对某些分类算法无法处理数值型属性提供属性预先离散化算法；针对挖掘前预先离散化所带来的偏差问题，在决策树算法基础上提出一边建树一边离散化的算法改进。 本文针对上述研究内容，进行了反复的研究与论证，结果表明，本文的理论，方法与技术正确有效，为数据挖掘项目的开发提供了有效的理论指导，具有良好的实际应用前景。更多还原

【Abstract】 Nowadays the society is full of all kinds of information. It becomes more and more important to make the best use of the information. As a result, more and more people focus on data mining, as a newly-established frontier subject and the most active field in the research of information system and computer science.Guangdong Provincial Project esearch of Development Tools and Intelligent Analysis Platform on Data Warehouse (A1020103), sponsors the thesis. Based on exploring and analysis on the related literatures of data mining theories and applications, the thesis summarizes the main research contents and key technologies in this field, commenting the development of trends, questions and further tasks on data mining system, and studies some questions in the process of data mining: data mining project development process model, data quality analysis in the process of data preprocess, pre-discretizing numeric attributes algorithm and improving the decision algorithm on dealing with numeric attributes based on the data warehouse(provided by Visionsky Company) of Bank of China. The main works developed in thesis as follows:1> The thesis studies the related theories, applications and trends around data mining system, such as data mining technologies, data mining methodology and data mining system.2, Aiming at the question of ignoring project layout, the thesis brings forward new data mining system development process model based on project at the existed data mining process model, such as CRISP-DM process. There are four stages in the new process model which are project layout, mining preparing, mining and project evaluating. The new process model emphasizes on describing the project layout phrase to make data mining technologies feed on business application.3 ?Guided by the new process model, Credit Card Mining System was successfully designed, the thesis provides the solutions about the key technology questions in the Credit Card Mining System based on classification technology: designing the standards of data quality in the data preparing phrase, pre-discretizing numeric attributes algorithm and improving the decision algorithm on dealing with numeric attributes.We did a lot of experiment and evaluation to proposed the process and the algorithm. The experiment results are encouraging and show that the theory and method presented in this thesis are correct, efficient, valuable and a good guider in practical application.更多还原