The Research And Application Of Bi System In Retail Industry

After China entered into WTO, with the pace of opening to the outside world quickening; there will be more intense competition in the field of trade enterprise, especially the retail trade industry will face an unprecedented degree of challenge. In the future, if the enterprise wants to survive himself in the intense competition, he has to be able to look over the history and fasts to see clearly the future tendency, thus to make the decision-making fleetly and to face the changes of the market. The fast decision-making comes from the correct and plenty analysis to the market, customers, the enterprise marketing condition and so on. However, as a manage system, the intrinsic ERP system can not be competent for this task, while the Business Intelligence (BI) will do it well. BI is a new technology, which makes integrative use of Data Warehouse (DW), On-Line Analytical Processing (OLAP), and Data Mining (DM) and so on. Via BI, we can transform raw data into usable information and knowledge.This paper is base on the analysis platform project retail trade oriented which sponsored by the electronic information industrial development fund, proposes BI system's overall construction through analysis and research of data warehouse, ELT and data mining technologies.By using the B/S structure for development, this system reduces the development cycle and easy for implementation. Complied with the MVC pattern for development, we design the presentation layer, service layer and data persistence layer, which make the structure of software much clearer, increase the expansibility of the system and easy for upgrade and management.This paper also analyzes the classical Apriori algorithm of rule mining algorithms. This algorithm will take a long time and large storage space when analyzing large-scale data. In view of this shortcoming, this article proposes the Apriori algorithm which is based on matrix. Because of the disadvantages, this paper proposed a new Apriori algorithm based on the matrix, and we apply the new algorithm to the market basket data analysis successfully. It will supply the effective decision-making supporting information for the superintendent, through the discovery of the customer purchase model.