The Construction Of Business Analysis System Of China-Unicom And Analysis Of Product

With the opening of the market of telecommunication, the competition between the telecommunication operators will be fiercer and fiercer, the reducing of the profit makes every operator change from extensive management to intensive management, in order to meet the fierce competitive environment of market day by day, give play to one's own comprehensive advantage, realize the flexible and changeable marketing tactics, offer satisfaction with more pertinence service for customer; Dispose and optimize one's own resources rationally to the maximum extent at the same time, reduce the operation cost, strengthen key competitiveness, need, set up one include datum warehouse and data analysis one facing business analysis system that decision analyzes intact of technology as soon as possible.With Business Analysis System (BAS) of the Tianjin Union Communication company as the research background, and elaborates the building of data warehouse system from data warehouse and OLAP the two part, especially told the design and operation process of ETL part. The system helps the analysts to get business information effectively and timely from huge data source. It has steadily effect after being used, this is a good model for the design of likely system. And according to package service problem of Union Communication vocation, this thesis adopts clustering analysis and association analysis technique to set up package analysis model. Firstly, the article deeply analyzes the clustering analysis and association analysis technique, afterwards the article brings forward a package analysis model based on Apriori algorithm and k-means algorithm, and narrates the process of model establish in detail. Aiming at slow speed problem in processing because of large data set, this paper improves on the Apriori and k-means algorithm, increases model capability. Finally, the article gives the test for the model. The test result demonstrates that the model comes to a higher accuracy and can react customers' consume habit. It indicates that the design of package analysis model is successful and the model comes to the request of business usage.