Data-aware Applications, M-zone Brand Customer Classification

M-Zone's target user groups are pre-paid telecom operators an important part of the user, characterized by young, fashionable, trendy, personalized, and more emphasis on text messaging, The internet and other data services used. Seize this part of the young, potential customers, the telecom operators to seize the future potential of high-value customers. They mainly through sub-goals, and provide personalized customer service to develop marketing strategies, the combination of a tariff for their business and the implementation of differentiated services, not to characteristics of product features but by the customer market segments. How different segments of the market for college marketing development for the telecom operators more important.This thesis focuses on these issues, using data mining technology in M-Zone target user group segmentation campus users. By comparing the current mainstream of several classification methods, marketing and decision-making process based on the data size, computational accuracy and efficiency requirements, choose to MLP-based, with local parameter adjustments to optimize the way, using distance-density method clustering method, to achieve the M-Zone campus recognition and classification of user groups, and conducted the actual verification of several examples of marketing. The main contents are:1. Analyzed and compared in detail the current mainstream of several classification methods, including Bayesian classification, decision tree, perception classification, artificial neural network classifier, KNN classification, and various methods for the environment;2. On a detailed analysis of needs, according to user groups from the identification, rational application of resources, new product research and development point of view, combined with the dynamic classification of users, modular and rapid classification and clustering requirements, select the classification method;3. Detailed design of the system model, the deployment of the entire project implementation steps, including a modular design, sample preparation, data preprocessing, sensor generates and optimization, user clustering, the actual user authentication, etc.;4. Perception algorithm and the optimization of the process and implementation of distance-density clustering algorithm was introduced;5. The actual user identification and implementation of marketing activities, a summary of results achieved the purpose of this research, and proposed several ways for further improvement.