一个基于Web的金融多层聚类系统  

A Financial Multi-Level Clustering Analysing System Based on Web

作　　者：梅永国[1] 杨杰[1] 陈文中[2] 

机构地区：[1]上海交通大学图像处理与模式识别研究所,上海200030 [2]香港理工大学计算机系

出　　处：《上海交通大学学报》2002年第12期1816-1820,共5页Journal of Shanghai Jiaotong University

摘　　要：介绍了一个基于 Web的用于金融数据挖掘的多层聚类系统的设计与实现 .该系统基于金融比率对香港、大陆股市中的上市公司进行了聚类分析 .它采用 3层的体系结构 ,即用户层、应用层和数据库层 .应用层使用了 Java Servlet、Java Script和 JDBC等编程技术 .系统数据库中存储了大量的金融数据 ,并实现了 2 4个金融比率的计算和存储 ,用户还可以自己定义新的金融比率和众多金融比率之间构成层次关系 .用户可以在不同的比率层次上对公司聚类 ,显然 ,采用多层次的聚类比单层次聚类有明显的优越性 .测试结果表明 ,该系统灵活、快速、扩展性好、结果易理解 ,有助于金融专家和有经验的投资者进行正确分析和合理投资 .This paper introduced a financial multi-level clustering analysis system, which, based on the historical financial ratios, applies the cluster analysis technology to analyze the listed and non-listed enterprises in China and Hong Kong. The system adopts a 3-tier architecture, i.e., user tier, application tier and database tier. JavaScript, Java Servlet and JDBC are used in the middle tier server. There is a huge financial database, which implements 24 financial ratios, in this system. Users can also define new ratios, and all these ratios have multi-level relations among themselves. Users can do clustering analyses on different ratio level. Apparently, multi-level clustering has advantages over single-level clustering. The test result shows that this system has many good characters such as flexibility, swiftness and scalability. Thus, this system can benefit to the financial analysts and the experienced investors for their own particular analysis purpose.

关 键 词：金融多层聚类系统 数据挖掘 聚类分析 金融比率 WEB JAVASERVLET JDBC DATABASE 

分 类 号：F830.49[经济管理—金融学] TP39[自动化与计算机技术—计算机应用技术]