数据[资料]数据仓库论文：河北网通基于数据仓库的帐务决策剖析系统的设计与完成

数据仓库论文河北网通基于数据仓库的帐务决策分析系统的设计与实现

【中文摘要】数据仓库技术是基于数学及统计学严谨逻辑思维的并达成“科学的判断、有效的行为”的一个工具。数据仓库技术也是一种达成“数据整合、知识管理”的有效手段。它从不同视角对企业经营各方面的信息进行精确分析,准确判断,为企业的决策者做出正确的决策提供了重要的支撑。典型的数据仓库系统,比如:经营分析系统,决策支持系统等等。也随着数据仓库系统带来的良好效果,各行各业的单位,已经能很好的接受“整合数据,从数据中找知识,运用数据知识、用数据说话”等新的关系到改良生产活动各环节、提高生产效率、发展生产力的理念。本文结合已有的数据仓库技术理论,对河北网通的基于数据仓库的账务决策分析系统项目进行了设计与实施。提出了面向数据仓库的账务决策分析的体系结构及相关主题域,并根据“从一个简明主题开始实施”的原则,采用自底向上的方法进行构建,首先建立部门数据集市,实现区域自治,对相关部门的应用进行分类复制,对每个数据集市进行数据重构,逐步扩展到企业数据仓库(EDW) 。重点工作是在账务主题域的数据集市构建过程中实现了客户、账户、用户实体,帐单、欠费、存款、清单、产品、优惠实体等主题域信息,根据不同分析需求实现不同汇总粒度的划分和数据的分割,实现了不同数据源数据的集成等,并针对收益分析构造了多维模型,基于此模型利用固定(预定义)报表、图表、即席查询(Ad-Hoc) 、

EIS(Executive Information System) 、多维动态分析、数据挖掘等展现方式,从不同维度、不同粒度、多种视角形象直观地显示分析数据。重点对计费系统账单数据进行OLAP分析,建立了面向计费系统的财务决策的数据集市。通过该项设计与实施,探索出了一种将数据仓库技术与具体应用相结合的有效的方法,以业务支撑系统的数据以及其他相关数据作为基础数据源,采用科学的数据抽取、整理、存储等方法,挖掘数据内部蕴藏的大量有用信息,通过科学的分析,做出可信的预测,从而为市场决策管理者和市场经营工作提供及时、准确、科学的辅助决策依据。本项应用为提高公司整体管理效率,确定决策分析方向提供了数据依据。本文的经验可供其它数据仓库项实施借鉴,研究成果经进一步完善和扩充可推广到与通信行业相关的各企业应用。

【英文摘要】 Data warehouse technology is a tool based onrigorous mathematical and statistical logical thinking toreach “scientific judgment and effective behavior, ” but alsosome efficient approach to “data integration and knowledgemanagement. ” The technology conducts accurate analysis andjudgment from different perspectives so as to provide keysupport for decision-making. Typical data warehouse systemincludes business analysis system, decision support system,etc. With the good results displayed from data warehousesystems, the concept has been well-accepted among

organizations in various walks of life that “data integration,data application and data speak” do matter in the process ofproduction improvement, and help develop productionefficiency.Combining the existing data warehouse theory, thepaper designs and operates the data warehouse-based HebeiNetcom accounting decision-analysis project, and furtherproposes the data warehouse-oriented accountingdecision-analysis system structure and related domain fields.The system is started“froma simple domain”and set up accordingto bottom-up approach. The department data mart is firstestablished to achieve regional autonomy and classifiedcopying of department applications, followed by datarestructure of datamarts and further expansion into EnterpriseData Warehouse (EDW) .During the data mart building, the keytask lies in the realization of domain information, includingclient, account, entity, bil l, arrear, deposit, inventory,product and offered target. Analysis of various demand leadsto different aggregate granule partition lattice and datacategorization, achieving the data integration from varioussources. Moreover, multi-dimension model is constructed forbenefit analysis. By using fixed (predefined) reports, charts,ad hoc query (Ad-Hoc) , EIS (Executive Information System) ,

multi-dimensional dynamic analysis, data mining and otherdisplaying methods, the model achieves the visual datademonstration and analysis from different dimensions, granulesizes andperspectives. The billing data is especially focusedon for OLAP analysis so as to establish a datamart for financialdecision-making in billing system.By designing andimplementing the project, an effective approach has beenexplored to combine data warehouse technology and specificapplications. Taking data fromoperation-supportive systemandother relevant systems as the underlying data source, theapproach employs scientific data extraction, collation andstorage methods to dig out a lot of useful information, andmakes reliable predictions after scientific analysis, servingas timely, accurate and reasonable support for decision-makingand market operation.The application of the project providesdata support for companies to improve overall managementefficiency and determine direction of decisions. Theexperience of the paper could serve as experiment for other datawarehouse projects and be applied to telecommunication-relatedenterprises after further perfection and expansion.

【关键词】数据仓库数据集市OLA P粒度划分EDW

【英文关键词】 Data warehouse data mart OLAPGranule Partition Lattice EDW