Behaviour Rule Prediction Of Mobile Communication Users And Development Of Data Mining Platform

With the development of information technology and mobile internet,more and more companies are generating huge amounts of data in their operation,data mining technology is also becoming increasingly mature.Companies are increasingly paying more attention to turning these data into useful information and knowledge through data-mining techniques to guide the enterprise operation.The rapid development of mobile Internet has brought great challenges to our telecom enterprises.But mass data also bring new growth points.The main content of this thesis is about how to bring data mining technology into the operation of telecom operators to discover user behavior patterns and hidden useful information from massive data.It can provide useful information for market,operation and technical implementation to achieve the purpose of saving company operating cost,improving economic efficiency and realizing intelligent operation.Firstly,the thesis introduced the behavior rules prediction and the development of data mining platform based on mobile communication user’s data.We introduced the basic flow of data mining technology and the basic principle of classification algorithm and then analyzed the application scenarios of data mining technology in telecom enterprises.In the part of platform realization,we analyzed the functional requirements with the consideration of characteristics of data and the actual demand of enterprise operation,and designed the functional framework of the platform and the specific operation process.Then the implementation of each function module is introduced including data inputting and preprocessing,model construction based on different classification algorithms,model visualization,model evaluation based on ROC curve,application of model.At the part of model construction,the principles,processes,and implementations of several common classification prediction algorithms are introduced in detail,including CART decision tree based on Gini coefficient,CART decision tree based on information gain,nearest neighbor algorithm,Bagging and AdaBoost integration algorithm based on decision tree.The platform meets the complete flow of data mining,it can construct classification prediction model on user’s historical behavior data to analysis user behavior rules and guide telecom enterprises to conduct customer relationship management and business operation.Then,the practical applications based on mobile communication user behavior data and data mining platform are introduced.The part introduced how to find target customers in flow promotion and terminal marketing by data mining technology.In the two practical applications,data acquisition,data preprocessing,model construction,model evaluation and application are introduced in detail.Finally the thesis analyzed the value of data mining for the operation of enterprises.The robustness and availability of the platform are tested through practical analysis,and a new solution are provided for the management of customers and company operating in telecom enterprise.