Study And Practise Business Intelligence System Based On Data Warehouse

Business Intelligence(BI), is a technology which transform the data store in all kinds of information system into useful information. It's usually regarded as a tool for enterprise to transform the resent data to knowledge and make a wise management decision. BI plays a rital role during the growing process of the enterprises, because it help the enterprise accurately master the varying business environment and make rapid and accurate management policy. Data Warehouse, Online Analytical Processing and Data Mining consistute the supporting technology for BI.Above all this paper gives an introduction of the relavant content and technology of the BI system, then designs the BI system construction based on the B/S mode of Web, including the DB layer, the DW layer, the OLAP&DM application services layer and the data browsing for terminal users layer. In this construction , the layers are mutual independence so that they have more flexibility and adaptability to the changing damand of end users. The leading users can browse the data by IE, which make easy for non-professionals to operate, as well as for administrators to update and maintain the system, and with a full usage of enterprise's present LAN to reduce the invest.Follow it , the paper analysises the data construction of marcket's original information system, according to the actual demand, designs the three themes of DW: the saling information analysis, the unit price analysis and the vegetable detecting analysis, then builds the DW in the SQL Server 2000 based on the three themes, and extracts, transforms and loads the relevant data from DB to DW. Then it establishes three correspondent multidimension data set for every analysis theme with the Analysis Services of Microsoft ,use MDX to realize not only all kinds of analysis demand but also the analytical method such as roll up and drill down, slice and dice ,etc. It also developes the web-based OLAP exhibition with Microsoft's Reporting Services and discusses the data security of terminal users. Finally, on the foundation of analysisingone of the data-mining algorithms------Decision Tree, the paper designs two predicatingmining models with Microsoft_Decision_Trees, individually mining and predict the contribution of product owners to the sales and unit price difference ratio.In conclusion, the paper puts forward two future research direction: construct the analysis models of business management and establish the information resource plan of management information systems.