Research And Implementation Of User Behavior And Phone Replacement Prediction Technology In Mobile Internet

With the rapid development of communication network and the widespread popularity of intelligent mobile terminals,the mobile Internet has also developed rapidly.It has not only provided a lot of conveniences for people's lives,but also brought new challenges and opportunities for the operators and related enterprises.On the one hand,more and more users access the Internet through the mobile terminals.Considering that the amount of network data and types of services have soared,users' needs have been submerged in the massive resources of the mobile Internet.So how to dig out users' behavioral characteristics and personalized needs is an urgent problem for the operators and enterprises.On the other hand,using the DPI business identification technology and data mining technology to analyze user preferences and predict potential terminals replacement users has become an important breakthrough for operators and companies to increase market competitiveness and business growth.So this thesis develops research on technology of mobile Internet user behavior analysis and terminal replacement prediction.The main work of this thesis is as follows:1.Based on the deep research of signal analysis,web crawler,DPI,data analysis and data mining technology,a system of mobile Internet user behavior and terminal replacement prediction was designed.The system combines hierarchical with modular design ideas,which can reduce coupling and improve performance,and has high practical value and significance.2.Using the multi-threaded topic web crawler to obtain Internet resources and store them into the database.Combined with the crawler information database,basic feature database and matching regular database,DPI technology is applied to identify specific contents of users' business.The recognition accuracy rate reached 90%.3.In the process of user behavior analysis,this thesis used the coefficient of variation to calculate the index weight,and then evaluate the users' behavior objectively through the weighted multi-index evaluation method to deeply mine users' preferences.4.In the research of terminal replacement prediction model,the author conducted the comparison of logistic regression,neural network and CHAID algorithm,and then selected logistic regression algorithm to realize the implementation of terminal replacement prediction.The terminal replacement prediction results were verified and can provide the guage for precise marketing and improve enterprise profit.This system can identify business on massive user data,analyze users' preferences,and predict potential terminal replacement users,which is able to provide support for personalized precise marketing by operators,Internet service providers,and mobile phone manufacturers.