Improving the Accuracy of Network Intrusion Detection with Causal Machine Learning

Table 8

Comparison of accuracy between CFS and CRFS in the CICIDS19 dataset.

Feature selection method	Accuracy under different types of cyberattacks (1, 3, 7, 11)
	1	3	7	11
KNN-CFS	0.9985	0.9953	0.9735	0.8917
KNN-CFS (do)	0.9981	0.9953	0.9715	0.8895
Percentage	0.9995	1	0.9979	0.9975