Authors: Arie Segev, Shin-Chung Shao, J. Leon Zhao
Addresses: Haas School of Business, University of California, Berkeley, California, USA. Jason Technology Ltd., Hsi-Chih, Taipei, Taiwan. Department of MIS, University of Arizona, Tucson, Arizona, USA
Abstract: Business intelligence (BI) has become an integral part of e-business activities of companies. The success of such effort is dependent on integration and management of terabytes of transactional data in a large-scale data warehouse to support various analyses. In the last two years, web-oriented BI and outsourcing have led to more challenging problems of integrating these new types of data with the internal transactional data. Furthermore, recent trends in enabling the real-time company have placed new performance demands on BI software. Significant research has been carried out in optimising multidimensional aggregation in data warehousing that serves as the base data for BI analysis software. However, virtually all of this work has focused on either simple aggregates or on specialised analysis such as in human genome and other scientific applications. In this paper, we focus on statistical analysis models and techniques to reduce the volume of base data and enable real-time BI analysis. We introduce a new data aggregation model, referred to as the multivariate and multidimensional aggregated data model (M²AD), for supporting statistical computing in a data warehouse environment.
Keywords: business intelligence; data aggregation; data mining; data warehouse; decision making; statistical analysis; summary data.
International Journal of Internet and Enterprise Management, 2003 Vol.1 No.1, pp.7-30
Published online: 18 Jul 2003 *Full-text access for editors Access for subscribers Purchase this article Comment on this article