Study On User Behavior Preferenceof Mobile Internetbased On Support Vector Machine

In recent years, the rapid development of mobile Internet has become one of the most remarkable phenomena in the field of world’s information and communication, and the number of users is rapidly growing, and which promotes the continuous innovation boom in the current mobile Internet field. With the fast development of Mobile Internet, the users’preferences is more diverse, and user behavior is also showing new features, so it brings great opportunities for the telecommunications industry. The mobile internet user has not only always been the core of research, it but also is the source of sustainable development and competitiveness in the mobile Internet enterprise and service provider. How to explore and understand the user’s behavioral preferences and provide personalized service is an important issue of businesses willing to invest capital and research.With the development of computer technology and the database, people began to use advanced computing to analyze user behavior, in which support vector machine approach is increasingly applied to various fields and get the attention of scholars in many fields. Based on the unique advantages of Support vector machine in solving the problems of nonlinear and high dimensional pattern recognition, it can be widely used. In the paper, the support vector machine theory is introduced to study the behavior of mobile Internet users in the field, and it effectively solves the defects of small sample of empirical research, and points out new methods and ideas for the research of mobile Internet user behavior, in addition, the support vector machine also expand a new application field.Methods this research use the literature research and empirical research, qualitative research and quantitative research, empirical data and the simulation and the combination of the traditional statistical analysis and data mining by combining modern, and the key factors affecting the mobile Internet user behavior preference and mobile Internet behavior preference are studied.The main outcomes of this study are completed as follows:1.It reviews and summarizes the support vector machine theory and discrete choice model, and discussed the relevant research and development status of mobile Internet in detail; and analyzes the basic theories and characteristics of statistical learning theory and support vector machine.2.From the perspective of020, based on the two angles of Online and Offline to analyze mobile Internet user behavior preference. First, from the perspective of Online, it studys mobile Internet user behavior preference in mobile phone online, and mobile phone news client as an example, it studys user browsing patter, and through the analysis of the data, we found that the time distribution of mobile Internet case browsing news is a fat tail effect. Fragmentation of reading has become one of the main characteristics of mobile Internet user browsing patterns.From the perspective of Offline to study user behavior, firstly it carries on the segmentation to the user, and analyzes the nature, user behavior analysis, perceptual attributes to select the main variables to study, and then through the telephone survey and operators system to obtain the corresponding data and begin basic statistical analysis, and it combined with the discrete choice model, using SAS tools to establish two element logit model, at the same time, it analyzed the selected variables to find the influence factors and obtaine the corresponding conclusion, to give the specific steps of the algorithm and empirical analysis, and finally gives some suggestions.3.Based on the influencing factors, it uses support vector machine to classify mobile Internet user behavior preference, and uses the MATLAB software and C-SVM two element classification machine, with RBF as the kernel function to analyze the model and use cross validation methods to find the optimization of parameters of the model and kernel function; and after training the function, comrrect rate (?)was catisf(?) it compared the clsssification accuracy of different kerrei functions and parameters. The results show that the method is the best, and use SPSS MODELER to analyze the importance of each predictor variables, and this study shows that the research method of support vector machine is suitable for mobile Internet user behavior. Finally, divide by six dimensions, paper divides into four groups of users, business people and white-collar focus on efficiency, migrant workers, service staff concerned about the price, student groups concerned about the adequacy of the phone traffic packages, as well as the boss pursuit user experience, each class groups have shown their unique preferences and behavior patterns, and we propose corresponding strategies to guide the development of appropriate business strategy, which has important guiding significance in the future mobile Internet user behavior research.