数据仓库与数据挖掘在证券业中的研究应用

【作者】 徐峰；

【导师】 戚桂杰；

【作者基本信息】 山东大学 ， 管理科学与工程， 2005， 硕士

【摘要】 数据仓库是面向主题的、集成的、非易失的,是随时间变化的数据集合,用来支持管理决策。它是在传统的业务数据库的基础上,经过对大量静态业务数据库的提取,形成数据仓库数据。数据仓库可以为决策支持系统的决策支持处理过程提供有效的支持,也同样可以为其他的信息处理系统服务。数据挖掘可以称为数据库中的知识发现,它是从大量数据中发现并提取隐藏在其中的可信的、新颖的、有效的并能被人理解的模式的高级处理过程。数据挖掘解决了传统分析方法的不足,并能够对大规模数据的进行分析处理。数据挖掘从大量数据中提取出隐藏在数据之后的有用的信息,为人们的正确决策提供了很大的帮助。数据仓库、联机分析处理和数据挖掘三个技术在国外已经取得了广泛的应用,而在我国的应用才刚刚起步。证券业作为我国较早应用信息技术的行业,己经建立了较完善的事务处理系统。多年的应用也使得证券公司积累了大量的数据,其中隐含着大量有价值的信息。如何利用这些数据,深层次地挖掘信息,为公司的决策服务,己成为证券公司的当务之急。 我们通过查阅资料了解了国内部分行业内如金融、电信、零售业的企业在数据仓库和数据挖掘应用方面的现状,发现这些企业的大部分信息化基础建设已经做得比较好了,数据库中存储了大量的数据,但由于各种原因,如资金问题、专业技术人员等问题,在数据分析方面一直处在观望和准备的状态。本文分析了这种现状说明的问题和企业的困惑,针对这些应用上的问题和困惑,在较为详细介绍了数据仓库和数据挖掘工具方面的知识,如数据仓库的定义、结构,数据挖掘的定义、工具和应用过程后,以证券业的数据仓库和数据挖掘应用为研究对象,通过对它的分析和调查研究,探讨数据仓库的实际建立过程和数据挖掘的应用。包括从最初的各种需求分析(数据环境分析、业务数据库结构分析、应用系统主题分析)到数据仓库应用系统结构设计、数据模型设计,数据仓库的数据转移方案、数据加载、创建多维数据集,最后利用数据挖掘软件和其他数据分析工具,对抽取的数据进行OLAP分析和股票行情关联分析等,找到理解分析结果的途径,使企业对于数据仓库与数据挖掘在企业中的应用有一个较为直观和详实的认识,更多还原

【Abstract】 A data warehouse is a subjected-oriented, integrated, time-variant, and nonvolatile collection of data in support of management’s decision making process. It is based in traditional business database, by withdrawing the large quantity business database, to form data warehouse data. The data warehouse can support for the decision-making of the decision support system and handle for the other information service system. The data mining is also called the knowledge discovery in database, it discovers from large quantity of data and find authentic, novel and effective model that can be comprehended by people. The data mining finds useful information from the data of large quantity to support for reasonable decision-making. The data warehouse, online analysis and data mining have obtained the extensive application abroad, but in China the application just starts. The stock industries use the information technique early in China and establish perfect OLTP. The application of many years also makes stock companies backup flood of data, among them implicit worthy information in large quantity. How to make use of these data, discover information on the deep level and serve for the management’s decision making, becomes the urgent matter of the moment of the stock company.We look up to the files to understood the present condition of local parts of industries, such as the finance, telecommunication, retail trade, in data warehouse and data mining. We discover that majority of f these enterprises have built perfect IT infrastructures, but because of some reasons, such as funds, profession technical personnel, they make few development in data analyzing. The paper analyses the explanatory problem in this kind of present condition and the perplexity of the business enterprise, aiming at these problems on the applications. We introduce the knowledge about data warehouse and data mining in detail, such as the definition and tools of data warehouse and data mining. We take the application of data warehouse and data mining in stock as example. By investigating and anglicizing in it, we canlearn the process of establishing data warehouse and applying data mining. The paper includes the requirement analysis (data environment analysis, structure analysis of database and subject analysis of system), the data warehouse application system construction design, data model design; transferring data in data warehouse; data loading and creation of multi-dimensional data sets. Finally we use data-mining tools to make OLAP analysis and association rules analysis on stock data warehouse to find ways to comprehend outcome of analysis. The enterprises may have a direct and thorough view on application of data warehouse and data mining in business and how to realize it in their own business.The paper introduces the latest development and basic principle of date warehouse and data-mining, discussing the technique in data warehouse application meaning in stock company. The paper discuss the theory and ways of establishing data warehouse based on business system foundation in stock company, providing the basic theories for the data warehouse application in stock company and practical methods. Required with the business and data characteristics of the stock company, the paper puts forward the total frame of the stock company data warehouse system. Around the stock sales subject, we set up a case about design of data warehouse to find how to design target data warehouse in stock company. The paper discusses ETL techniques which transfer the data to data warehouse., and complete the transformation by the help of MS SQL Server to built multi-dimensional data sets about the stock sales.The paper is divided into totally six chapters. Chapter 1 introduces the data background and the meaning of data mining; Chapter 2 explains the concerning basic theories upon the data warehouse and data mining; Chapter 3 analysis the present progress made by researchers in and abroad; Chapter 4 expatiate meaning of establishing data ware house in stock industry and process in detail; Chapter 5 expatiate that we proceed the OLAP analysis and association rules analysis on the basis of data warehouse; Chapter 6 is the summary of total paper.The innovation of paper is that we construct two stock sequence rule model with time constraint: the stock’s sequence rule model of one dimension with certain timesegment (represented by w) constraint and the stock’s sequence rule model of two dimensions with W and time-interval (represented by INT) constraint.The innovation of paper is that that we construct two stock sequence rule models with time constraint: the stock’ sequence rule model of one dimension with certain time-segment (represented by constraint) and the stock1 sequence rule model of two dimensions with W and time-interval (represented by INT) constraints. This means that if the closing price of stock A is going up to X% in time-segment W, then those of stock B and will also rise (or descent) in Y% probability in time-segment just after INT time-segments. By this we can find some useful model that can hardly be found with traditional statistical methods.In chapter 5 we make a substantial evidence to verify the possibility of model presented in paper. In the stock analysis realm, this model will help the investor make more reasonable and more complete analysis, thereby improving the quality of the decision-making. This model also can be used in other realms such as bank, telecommunication industry etc.更多还原