Prediction And Analysis Of China Mobile Customers Based On Data Mining

The competition from both inside and outside the mobile communications industry, the China Mobile faced with, is becoming increasingly fierce. To improve its own competitiveness, China Mobile has to resolve following problems that not only ensure customers receive and handle new business, but also ensure no loss of existing customers. China Mobile had launched "4G services" at the end of 2013, In this paper, to meet the main demand of China Mobile that predict Mobile 4G potential customers, based on analysis of the raw client data provided by China Mobile, the following tasks have been completed:Firstly, basic theory and development trend of classification, clustering, and association in data mining were summarized based on relevant researches at home and abroad on the present stage of data mining used in customer information; Secondly, 4G potential customers were predicted by using Bayesian network, C4.5 and CART separately, and their classification models and predictions were comparatively analyzed; Thirdly, Mobile customers were finely divided into three clusters by using clustering algorithm, then, 4G potential customers of the cluster,in which customers have the potential to handle 4G services, were predicted by using the improved algorithm that was a combination algorithm of C4.5 and clustering algorithm. The improved algorithm raises the accuracy of the predictions and increases the number of 4G potential customers from 206 to 1649. Finally, lost customers of the cluster, in which customers have no potential to handle 4G services, were predicted by using the improved algorithm, at the same time, potential relationship between attributes of the cluster,in which customers have handled 4G services, were found by using association algorithm.Comparative analysis of the predicted results of improved algorithm and practical application results indicates that 72% potential customers have indeed handled 4G services.