Application Of Data Mining In Telecom Value-added Services

As an "Application-Oriented" technology,Data Mining(DM) has been an international hot topic that causes wide concern in both academic and industrial field.Facing fierce challenges from both aboard and at home,more and more Telecom enterprises have planed to establish the "Customer-Oriented" management mode. Taking use of Data Mining technology to find potential and valuable rules is an important approach that can improve self-competence for telecom enterprises.Therefore, it has high theoretical significance and application value.With new Telecom products and services coming up continuously,how to increase the number of users and realize Precise Marketing has been an urgent requirement for Telecom enterprises.Focusing on the application of Data Mining in Telecom Value-added Services,this article puts emphasis on the process of building the Prediction System for New business through data analysis and applying this system in potential clients forecast.First,the article gives a brief description of Data Mining Theory and related algorithms.It makes a detailed comparison and analysis of Classification and Regression algorithm.Second,it discusses the importance of Data Mining technology for Telecom enterprises and the current situation of the application of Data Mining in Telecom Industry.Meanwhile,it takes Fetion,the important recommended business of some telecom operator,as the object to study its developing situation,promotion problem as well as the features of Fetion clients.Third,the background and construction demand of Telecom Management Analysis System are described and a detailed discussion of function and application scope for the Potential Client Prediction System is made in this paper.Finally,aiming at forecasting the potential clients of new business, this paper deeply describes the design and implementation of the Potential Client Prediction System for Fetion.Based on Special-to-General analysis method and with the help of Clementine(the DM tool developed by SPSS),the prediction system chooses CRISP-DM,including Business Understanding,Data Understanding,Data Preparation,Modeling,Evaluation and Development,as the main frame.During the modeling process,the advantages of C5.0 Decision Tree,CART and Logistic Regression algorithm are fully utilized,the accuracy of classification is effectively improved,the stability of the prediction model is verified,and the goal that applying the prediction system to identifying Fetion clients is realized.Based on actual project,this article realizes the task that utilizing Data Mining technology in business prediction and guiding Marketing decision-making,which shows great commercial value.The application result indicates that the prediction model is scientific and accords with reality basically.Besides,it can afford necessary forecast information for Marketing and Sales Department.So it is significant for user-prediction in business promotion or the urgent need of user expansion.