Analysis of Intrusion Detection System by Applying Machine Learning Using KNIME Tool

Intrusion detection helps a lot for information security which enables companies to identify possible security risks and breaches to take immediate action. For monitoring network traffic, the intrusion detection system (IDS) alerts users to any suspicious or malicious activity. In this research, we primarily addressed intrusion detection and classification using machine learning approaches in conjunction with the Konstanz Information Miner (KNIME) tool. Detection of intrusion is implemented using the selection of features which involves one variate and it ranks features based on their relevance using information gathered from the KNIME node. Intrusion detection system applications are helpful for analyzing network data and identifying strange patterns that may indicate security flaws. To prepare the data and extract features, the CICIDS2017 dataset is used. Using KNIME nodes, a variety of classifiers were investigated in order to improve performance and compare with other classifiers. Performance evaluation is done based on F-measure, recall, accuracy, and precision. Decision Trees’ accuracy is 96.04%, Random Forests is 99.70%, Naïve Bayes is 98.99%, and Gradient-Boosted Trees is 89.90%. Random Forest performs better compares to others.