Optimizing the Effectiveness of Intrusion Detection System by using Pearson Correlation and Tune Model Hyper Parameter on Microsoft Azure Platform

Abstract

© 2020, World Academy of Research in Science and Engineering. All rights reserved. This paper studies the effectiveness of implementing classifier algorithm and Pearson correlation for an intrusion detection system. The effectiveness of intrusion detection system is evaluated based on the accuracy, detection rate and false positive rate. This study has been implemented by using simulation in Microsoft Azure platform. The machine learning algorithm is applied together with filter-based feature selection which provides the classifier algorithm to ensure the quality of the NSL-KDD, KDD 99 and CICIDS 2017 dataset. In addition, the tune model hyper parameter has been applied to enhance the performance of the classifier algorithm. The findings show that the implementation of classifier algorithm and Pearson correlation for an intrusion detection system has been able to improve the effectiveness of intrusion detection system in terms of accuracy, detection rate and false positive rate. Results from this study are useful for designing an effective intrusion detection system in the future due to the advancement of network attacks that are growing rapidly. This study can be further extended and improved by investigating the effectiveness of intrusion detection system in real network environment.