Research And Application Of Data Mining Based On Data Warehouse Of Telecom Information Analysis System

With our government further opening up of telecom industry and making a series of restriction policy adjustment, keen competition for our country's telecom market is occurring. Meanwhile, consumers desire higher and higher QoS. In the face of rapid increase of the cost of our country's telecom enterprise, making the most use of current resources and establishments to reduce the corporation costs and improve benefits, under the precondition of how to meet consumers' requirements and how to provide high quality service, becomes an important factor for telecom corporations to obtain the superior competition power. Obviously, these requirements discussed above need a powerful data management serving as decision-foundation and decision-support of supervisor. However, for the sake of a rapid increase of user number and service catalogue, the difficulty in telecom data management enlarges. One side, because it needs reserve enough details for retrieve, online working system becomes more and more inflexible; on the other side, according to fixed and time-lapse report system, supervisor and decision-maker can only get limited business-related statistic analytic information, and yet can not get and solve user feedback in time so as to make effective decision-making strategy to adapt the severe competition. So, DW(Data Warehouse) system becomes a focus for telecom corporation to obtain decision-making strategy and own analytic predictive information. Whereas the requirement of these backgrounds discussed above, this thesis bring forward a telecom information analysis system based on DW in order to meet the new requirement of telecom enterprise decision-making.This thesis starts from the whole architecture of the system, gives the system's subject analysis, completes the model building of the DW based on subject analysis, and introduces well rounded DM(Data Mining) algorithm for the decision-making assistant information's mining.This thesis has following research:(1) Design the subject analysis of DW based telecom information analysis system and its model based on subject analysis.(2) As for the subject of consumer division of the mobile phone fees' trait, a decision tree algorithm from data mining tools is given and it is applied in the theme of consumer division of the mobile phone fees' trait.(3) As for the subject of association analysis of the mobile phone fees' behavior, Apriori algorithm and FP_growth algorithm are compared, FP_growth is adopted to finish the analysis and a reliable effective decision-supported result is obtained.(4) Based on the models of the subject analysis, ascertaining the value of customercredit grade in the customercredit grade analysis is emphasized. The thesis gives a complex evaluating system of telecom customercredit grade based on AHP and achieves the value of the customercredit grade.