Internet traffic management using naïve bayes classification and principal component analysis

Abstract

Internet facilities is one important part of the infrastructure of the campus at this time. Internet facility is a part of teaching and learning activities. Important part of the internet facility is the internet bandwidth, which is often deemed less bandwidth for certain majors at certain hours of lecture hours especially active. To overcome this there needs to be an analysis and classification of the internet traffic at each point where the distribution of bandwidth is done so that in the end can provide information that can support decision for internet traffic management. One algorithm for classification algorithms used are Naive Bayessian, in which the classficication process before the beginning of the internet bandwidth usage data that exists in one period will be collected to be input to the Naive Bayessian algorithm for the distribution of clusters on the use of existing bandwidth based applications that use the internet and network users. But the initial dataset that of the Naive Bayessian is not optimal yet, to optimized it, the feature from initial dataset need selected so that the result from Naive Bayessian classficication algorithm became more accurate. Results to be obtained from this study is the selection of data feature can improve classification and analysis of Internet traffic based on user applications and the amount of capacity used by the user, which information the classification results can be used to optimize internet bandwidth.