Research On Internet Behavior Analysis And Prediction Method Of Mobile Users

Global Internet is going through the PC-Internet shifting to mobile Internet. The mobile terminals and mobile phones have dominated the development of the Internet. The world will become a true information age.The latest data of China Internet Network Information Center CNNIC in 2015 shows that the scale of Chinese Internt users has reached 668 million until June 2015, of which the scale of mobile phone users has reached 594 million. Correspondingly, the proportion has declined by using desktop, notebook and ipad to access the Internet.Fast explosive growth of data brought great pressure to store, process and analyze data. The introduction of big data technology not only meets the needs of the system functionality and performance, reducing the cost of IT deployment, but also extends the application field of intelligent data analysis. Big data technology has become a powerful and strong tool that make enterprisre enhance competitiveness in the era of data explosion.The traditional user behavior analysis lacks of data mining and analysis means, no segmentation the needs of user and position accuracy, which leading corporate market and user need separated. OTT business gave operators squeeze that results the huge loss of SMS, MMS, video and other services. Operators urgently need to build a Blue Ocean market based on big data.In this context, we propose an integrated approach and find an effective solution to analyze the user behavior of mobile Internet, which has become important issues that the mobile operators need urgently to resolve during the development of the mobile Internet.This thesis aims to implement precision marketing and researches the user behavior characteristics based on the Internet records. Combined with the user’age, gender and the user package, it builds a data analysis system that can classify and identify for mobile Internet behavior by using big data and data mining based on the mobile Internet. The data analysis system has implemented thematic analysis of big data, studied the model of customer segmentation and predicted the key issues of user behavior analysis.Research work of this thesis can be summarized as the following:(1) For the diversity and complexity of network user behavior, this thesis has studied the contents and methods of user behavior analysis, focusing on selection of data mining algorithms.(2) For the traditional behavior analysis lack of behavior analysis and data mining means, this thesis has used Hadoop, Hive and ZooKeeper to study the construction method of mobile Internet user behavior analysis and design the detailed module of data analysis layer, which is the foundation of behavior analysis.(3) Most clustering algorithms have better effect in dealing with low dimensional and a small amount of data. The quality of the clustering is reduced in processing the high dimensional and large data. According to the above problem, the thesis proposes an improved κ-means algorithm to solve the original algorithm in selecting k value and initializing cluster centers. A mobile user segmentation model is established on this basis.(4) The thesis takes C5.0 decision tree algorithm to identify clustering model and divide the subscriber’s home into group by analyzing the mobile user behavior. So it has been labeled and identified for users.(5) Decision tree algorithm is relatively good at selecting the most important variables, using the algorithm selected variables to build artificial neural network model and predict the user behavior.The results show that it is effective by introducing big data technology in the user behavior analysis system. It not only highlights the users’ portrait, but also achieves precision marketing.The thematic analysis of big data has become an inevitable trend of the industry development. It is of important role to grasp the user’habbit and preferences, position the Internet user accurately, mine data value and enhance enterprise competitiveness in the market by analyzing mobile user behavior to the Internet. The thesis has laid a solid foundation for big data and data mining applying in the telecommunications and provided a basis for making decision. It is of important significance not only theoretical research but also practical applications.