Machine Learning Technique for Classification of Internet Firewall Data Using RapidMiner

Abstract

A firewall system is a type of security system that controls how data packets enter and leave a network. It does this to improve cyber defence and decide what to do with harmful packets. Traffic packets are checked against criteria to stop possible cyber threats from getting into the network. This study demonstrates the classification of internet firewall data using a public dataset containing 65,532 records and 11 features, and a unique machine-learning technique. No use has been made of any personally identifying information in the filtered data. The action characteristic is selected from among these features as the class label. The action class now supports the options “allow,” “deny,” “drop,” and “reset-both.” The study proposes an intelligent classification model that firewall systems can use to determine what to do with each received packet using a machine learning algorithm and the RapidMiner tool to look at a packet’s properties. We classified the data using the Decision Tree (DT) and K-Nearest Neighbor (K-NN) methods. The highest accuracy was achieved using the Decision Tree model.