Issues Related to Sampling Techniques for Network Traffic Dataset

Network traffic data is huge, varying and imbalanced because various classes are not equally distributed. Machine learning (ML) algorithms for traffic analysis uses the samples from this data to recommend the actions to be taken by the network administrators. Due to imbalances in dataset, machine learning algorithms may give biased or false results leading to serious degradation in performance of these algorithms. Since the network dataset is huge, during training machine learning algorithm takes more time and hence sampling should be used to reduce the training time. But using sampling may cause loss of information which should be taken care off while obtaining the samples. In this paper various sampling techniques have been analysed for loss of information and imbalances during sampling of network traffic data. Data set is collected from the Panjab University network. Various parameters like missing classes in samples, probability of sampling of the different instances have been considered for comparison.