Research On The Prediction Of Unpacking Of Mobile Marketing Terminals Based On Data Mining

In recent years, the market of telecom industry will tend to saturate gradually because of the rapid development. For the limited customer resources, China Mobile, China Telecom and China Unicom are in fierce competition. Meanwhile, with the arrival of3G and4G era, people’s demand for smartphones will grow increasing. Therefore, China Mobile Communication Corporation takes the strategy of smartphones terminal marketing, in order to maintain the stock market and open up new market.As the company’s marketing means, its effectiveness also relates to the importance of competitive advantages gained.Therefore, it is important that we should improve the quality of terminal marketing activities, prevent and eliminate the unpacking behavior, as well as choose the right mobile phones marketing. This paper focuses on how to predict unpacking customers based on data mining technology.Using data mining analysed the mass of the mobile customers, to explore the features of unpacking customers.Based on these, we can predict the customers who have the tendency of unpacking among all the customers participated in the marketing activities, then take precautions in advance.This is an effective way to improve the quality of marketing activities, and has great theoretical significance and application value.In this thesis, we designed and built integrated prediction models using customer data which come from Yunnan province Mobile Communications Corporation, including basic information, communication behavior and spending behavior.The model is built under the process of the CRISP-DM, using the SPSS Clementine12.0. Effective data cleaning, integration, conversion and exploration were made in the process of modeling to improve the quality of the data. Finally, three models such as decision trees, neural networks and logistic regression were established.Through training, assessment and verification, we select the optimal model.According to the customers list which the model predicted, combined with customer value, we can work on different precautions towards different customers, and take targeted marketing strategies to provide decision support, so as to improve the quality of terminal marketing activities.