Analysis Of Internet Behavior Based On Clustering Algorithm

With the rapid development of Internet technology and gradual promotion of educational informatization,campus network as an important platform for teachers and students for teaching,researching and communication,has become an important index to measure the modernization and information of campus construction.However,with the continuous expansion of the campus network scale and the increasing number of network users,it also brings some negative effects.For example,some students get addicted to the network heavily,spending lots of time and energy and occupying most of the bandwidth.Therefore,analysis of campus network user behavior is of great significance.For network administrators,they not only can better understand the overall operation of the campus network,but also can provide a more powerful basis for rational network planning;For school administrators,they can understand students' Internet habits and their dependence on the Internet,so as to prevent students from indulging in the Internet and affecting their studies.This dissertation is based on the campus network of Yan'an University and the data source is the database of user records.We take one month of data as the research object and complete the following work on the platform of Microsoft SQL Server 2012:(1)Data preprocessing: the data is pretreated with the SSIS tool;(2)The model of online behavior analysis is established: the classical K-means clustering algorithm is selected,and the parameters of the online behavior analysis model(The selection of K in K-means clustering algorithm)are determined by comparing the four clustering results,the clustering results are analyzed,this dissertation analyzes the user group of abnormal Internet behavior using the drill function of the online behavior analysis model;(3)The verification application of the online behavior analysis model: the analysis model of online behavior is verified by taking another set of data.