Improved K-means Algorithm in User Behavior Analysis

College Network Center has accumulated large number of user´s online behavior data. We tried to analyze these data in order to grasp the user surfing habits and potential relationship, and then summarize an effective strategy for user online management. First of all, we taken a university as the example, and collected the user online data to extract related fields for sampling. In addition, we proposed a figure to present the number of Internet users in different time. After that, we use k-means algorithm to cluster the records, improved the algorithm by considering the element of user information. Finally, we apply the improved algorithm on a dataset and get the conclusion.