The Design And Realization Of Cluster Mining System Based On Data Warehouse And OLAP Technology

With the development of computer data acquisition tool as well as the relational database technology, at present various trades need to store the mass of data. The traditional data analysis methods have some difficulties in dealing with the mass data, resulting in more and more serious data disasters. Through the functions of simple inquiry and report generation, only the surface layer data information can be obtained. But the internal relation and connotative information of the data attributes can not be obtained, i.e. ,which leading to the waste of resource. Data warehouse, On-line analytical processing, Data mining technology have provided an effective way in solving this problem.Cluster analysis, as an important research subject of data mining field, is developing rapidly, which plays an important role in many practical applications. Under the background of the sales data of Willmart supermarket, and on the basis of others' study of cluster analysis, we design and implement a cluster mining system based on data warehouse and OLAP technology.Firstly, the paper introduces the research background, the concept of cluster analysis, the classification of cluster analysis methods, and several main cluster algorithms which are integrated in the system.Secondly, introduces establishing process of Willmart supermarket's data warehouse and OLAP cube, which is the study foundation of this paper.Thirdly, the paper discusses the design and realization of cluster mining system based on data warehouse and OLAP technology. This section is the emphasis of the whole paper. This system is composed of low-layer data warehouse, on-line analysis platform and cluster analysis subsystem. The paper focuses on the design and realization of cluster analysis subsystem and its application in the Willmart supermarket's data warehouse. The design of cluster analysis subsystem mainly consists of the design of the data preprocessor module, cluster analyzer as well as cluster analysis result feedback module. The paper elaborates the design and realization of this three modules. Then we carry on the cluster mining of Wilmart Supermarket's data, and provide some cluster analysis instances and results.Finally, we summarize the whole study work, and make further prospect to the work.