Performance Evaluation of Classification Algorithms for Intrusion Detection on NSL-KDD Using Rapid Miner

Abstract

The rapid advancement of the internet and its exponentially increasing usage has also exposed it to several vulnerabilities. Consequently, it has become an extremely important that can prevent network security issues. One of the most commonly implemented solutions is Intrusion Detection System (IDS) that can detect unusual attacks and unauthorized access to a secured network. In the past, several machine learning algorithms have been evaluated on the KDD intrusion dataset. However, this paper focuses on the implementation of the four machine learning algorithms: KNN, Random Forest, gradient boosted tree and decision tree. The models are also implemented through the Auto Model feature to determine its convenience. The results show that Gradient Boosted trees have achieved the highest accuracy (99.42%) in comparison to random forest algorithm that achieved the lowest accuracy (93.63%).