The Design And Research Of Packahe Recommend System Based On Random Forest And Bp Neural Networks

In recent years,China's national telecom business volume increased year by year,has reached about 1370 trillion yuan in 2016.With the increasing demand for mobile communication services,the increasingly fierce competition between the major operators.Telecom operators have PB level internal data,the traditional business model just passively accepts the user requirements and cannot get their individual needs.So,it's difficult to fulfill the user's complex and various demands at the market.In this case,how to establish an effective model,which can expect user's potential demands,to improve the user's satisfaction is a huge subject in Telecom operators.This paper first introduces the background and significance of this research,and then introduces the techniques of data collection from the aspects of bill collection?data collection and signal acquisition.And then introduces the analysis method of user behavior under large data.At the same time,this paper focuses on two large data analysis methods: random forest and Bp neural network.After that,it describes the research environment of this paper,and introduces the big data's application in a local operator both inside and outside.Finally,the paper introduces the design and implementation of package recommendation model.In this paper,we use the real user behavior data of the operator to model the random forest and Bp neural network respectively,so as to achieve the goal of predicting whether the user has the tendency to increase the package.Through constant debugging,the accuracy of the two models is more than 50%,for fill of commercial conditions.After comparing the results of the two algorithms with the precision,the recall rate and the running time,we can find out the algorithm which is more suitable for the operators.