Research On Mobile User Behavior Analysis Based On Machine Learning

In recent years,with the rapid development of mobile Internet,mobile Internet technology has penetrated into every field of people's life,and has become a part of people's daily life.Users will generate a lot of browsing data in the process of using the mobile Internet.These data contain a lot of user behavior information,and if you can exploit these data,it will be a valuable asset.How to make rational use of these data is an urgent problem for telecommunications operators.Machine learning can discover the potential laws between data and excavate valuable resources from the data.Therefore,machine learning is widely used in data analysis and data mining.This paper focuses on mobile user behavior analysis and machine learning.First,this paper studies how to use machine learning technology to analyze and excavate the data of mobile users,and proposes a K-means K value adaptive algorithm based on DPC(Density Peak Based Clustering)(called DPCK-Kmeans).The new algorithm selects the initial clustering center of the K-means algorithm through the DPC algorithm,which overcomes the randomness of the selection of the initial clustering center points of the traditional K-means algorithm.In addition,the new algorithm adopts the adaptive K-means clustering method,that is,the adaptive selection of the number of clustering is realized by merging the nearest two classes at each cluster until the merging condition is not satisfied.Secondly,this article will put forward DPCK-K-means clustering method in the UCI and artificial simulation analysis was made on the data set,the simulation results show that the proposed clustering method in the clustering accuracy and cluster effect has a bigger improvement.After that,the clustering method proposed in this paper is applied to the user behavior data of a city-level company of the operator,and analyzes the users' Internet time,business preference and traffic characteristics.The results show that the proposed clustering method can be applied to the real telecommunications business data and has practical application value.