Research And Application Of Clustering Algorithms In The Analysis Of The Behavior Of Campus Wireless Network User

With the rapid development of Internet,network has gradually become a way of life.People's interaction in the network make the Internet highly integrated with people and machines,so the network contains a large number of potential behavior information.The study of user behavior under campus network is of great significance to the construction of campus network,the attention of students' dynamic and network service.In the face of huge amounts of data,timely and accurately explore the behavioral pattern of data and finding the available rules are research hotspot in recent years.The clustering algorithm of data mining for behavioral analysis provides the effective methods.This paper mainly studies the fuzzy C-means clustering algorithm,self-organizing neural network algorithm and density peak clustering algorithm,then uses the campus wireless network data as data sources to improve the algorithm,and discusses its application of user behavior analysis in campus wireless network.Firstly,the learning interest of the campus wireless network users is analyzed in this paper,and then fuzzy C-means clustering algorithm combined with self-organizing neural network is proposed.The algorithm avoids the error caused by the initialization of the fuzzy C-means clustering algorithm,and also solves the problem of the low accuracy of the self-organizing neural network.The experimental results of data analysis show that the algorithm used in this paper can get the overall student learning interest,and also can effectively reduce the number of iterations,quickly reach convergence,improve the accuracy of clustering results,which has obvious advantages in behavior analysis.Then,this paper analyzes the abnormal behavior of the users in the campus wireless network,and proposes a density peak clustering algorithm based on trajectory segmentation.The algorithm preprocesses data through the trajectory segmentation algorithm based on speed and direction,which can avoid the problem of long trajectory detection.At the same time,the density peak clustering algorithm can be used to cluster non-spherical data considering the diversity of trajectory data distribution,which has better accuracy and stability.The experimental results show that compared with the K-means clustering algorithm and the hidden Markov model,the algorithm has higher accuracy rate and the running time does not significantly increase.Moreover,it can cluster data with arbitrary shape of distribution and less affected by abnormal points,which can easier to detect abnormal trajectory.Finally,this paper builds a large data analysis platform based on Hadoop distributed framework to deal with the problem that a single machine cannot efficiently deal with huge amounts of data,and the main module design scheme was expounded in detail.At the same time,the student user behavior analysis model is designed and the analysis algorithm is improved.Then,students' attendance is analyzed to get the students' learning interest and the abnormal trajectory is mined through several experiments carried out on preprocessed data of student users of wireless network,the performance of the algorithm this paper proposed is evaluated.The experimental results show that the proposed algorithm can not only run well in parallel distributed Hadoop framework,which can significantly reduce the running time and improve the operation efficiency,but also can accurately assess students' behavior,provide help for the management of colleges and universities by way of information,it provides a reference basis for solving similar problem.