組織內一個整合統計分析資料來源的模式

Abstract

統計分析可以廣泛的應用於組織各階層的管理活動上，並可協助管理者了解及掌握組織活動的狀況及特性。組織內部統計分析主要的資料來源有日常性作業資料庫、歷史資料庫、統計調查資料庫及多維度資料庫，如果直接由這些不同的統計資料來源個別提供統計資訊，會受到一些因素的限制，無法充分發揮組織內部統計資料所能提供統計資訊的潛能。針對這些存在的限制因素，本研究分析了各種統計資料來源的資料特性，發現這不同統計資料來源的資料特性及資料結構雖有差異，但不同的統計資料來源之間，其資料則是存有關係，如果能有系統的運用這資料間所存在的關係，將這些不同的統計資料來源整合起來，將可克服某些限制因素，提升統計資訊提供的能力。 本研究立基於統計調查資料庫的調查資料結構特性，運用資料倉儲建構概念及物件導向分析方法，提出了一個整合統計分析資料來源的架構，有系統的整合組織內部統計分析資料的來源。本架構首先搜集管理者的統計資訊需求，並利用物件導向分析方法，將這些統計物件整合，形成屬性分析用途之Metadata，據以整合不同來源的資料庫，而形成一個統計資料倉儲的模式，以更有效率的方式提供管理者決策支援資訊。本研究並以範例說明所提架構與程序的可行性。

Statistical information is vital for business analysis, and information sources which form the basis of such analysis may be obtained from daily operational databases, statistical databases, and multidimensional databases. However, extraction of the information separately from those databases, is limited by many problems such as access inefficiency, historical databases as a whole, and security of confidential data, etc. This paper proposes an architecture to integrate the three kinds of databases by developing a statistical data store for statistical analysis according to the features of the statistical database, object-oriented viewpoint, and data warehouse method. The architecture includes three processes. The first collects and analyzes statistical information requirements in terms of statistical objects, which are further organized into higher survey objects. The next process is the back-end process of developing statistical databases from the existing operational systems according to the objects. The multidimensional databases can then be established from the statistical databases. The final process is the front-end process of using various methods to extract the statistical information from the databases. An example is presented to illustrate the process. If the statistical data store can be developed in such a way, the commercial database systems, OLAP, powerful statistical packages, and self-developed methods, may all be used to extract statistical information, and the capacity to provide such information will be significantly improved.