dc.description.abstract | IPB internet services isimproving every year, both from the infrastructure and service aspects. One of the infrastructure improvements is internet bandwidth enhancement. However, research on user characterization of internet usage in IPB has never been done. This study uses k-Means algorithm to cluster IPB proxy server log's file to obtain the characteristics of internet users. The research evaluates all user's log files including student, educational staff, and lecturer. This study resulted in 16clusters calculated based on the number of hits on the 14 site categories. Clustering resultsshowedthatthere weresimilaritiesin theuse of the internetfrom student dorm.In each cluster, thetotalaverage percentage of the sites with Service and Entertainmentcategoriesareabove 24% and 33%.Bandwidth management is designed based on member and utilization of each cluster. | en |