Feature Selection Based on Cross-Correlation for the Intrusion Detection System

Table 4

Improvement percentage of the CCFS method over MIFS and CFS methods for DT, SVM, KNN, and ND classifiers on the NSL-KDD dataset.

NSL-KDD (20 features)
Classification method	CCFS improvement versus MIFS (%)				CCFS improvement versus CFA (%)
	Accuracy	Precision	Recall	F1-score	Accuracy	Precision	Recall	F1-score
DT	2.48	1.44	2.50	1.97	0.38	0.05	1.02	0.53
SVM	3.29	4.78	5.63	5.21	0.60	1.33	1.04	1.18
KNN	3.68	5.27	5.29	5.28	1.74	2.09	2.04	2.07
NB	4.03	3.36	3.36	3.36	1.83	2.14	1.20	1.67