Abstract
Supervised machine learning is a method to predict a class for labeled data, to improve different QoS metrics of several scopes such as educational, industrial and medical etc. This paper presents in-deep study focusing on four supervised classifiers were used widely to distinguish or categorize TCP/IP network traffic model and how they can be employed, these four are Naive Bayes, Probabilistic Neural Network, Support Vector Machine and C4.5 decision tree. The classifiers are compared with regard to three significant metrics namely classification accuracy, classification speed and memory consumption. The implementation results of simulation and comparisons show that C4.5 decision tree introduce best results with high accuracy up to 99.6% using the benchmark dataset consist of 24863 packets compared to the rest three tested classifiers.