CHI2CV: Feature Selection using Chi-Square with Cross-Validation for Intrusion Detection System

The internet technology has become essential needs among communities. However, this digital era threatens the security of important data and information. There are irresponsible parties that attempt to send intrusion attacks on computer networks. So, we need a security system, like Intrusion Detection System (IDS). Unfortunately, the attack types are getting more and more fast growing. There needs to be an effective intrusion detection model with a high degree of accuracy, namely by selecting the best features that can be used for the attack detection process. The feature selection method that we propose is the use of the chi-square test to determine which features are the most irrelevant to be discarded. Then from the remaining features, we need to choose the best features that can optimize the performance of attack detection on computer networks. For this reason, this research proposes a combination of chi-square and cross validation. The experimental results indicate that this proposed method has a significant impact on increasing the accuracy of detection of attacks on the network, from 95.51% to 96.70%.