Evolutionary Data Preprocessing to Alleviate Class Imbalance

Table 9

Experimental results of the original training, validation, and test datasets for each classification algorithm.

			Random subspace	Decision tree	SVM
Class	Normal		0.985	0.986	0.986
	Exploits		0.884	0.810	0.909
	Reconnaissance		0.693	0.691	0.647
	DoS		0.283	0.402	0.000
	Generic		0.003	0.981	0.972
	Shellcode		0.853	0.756	0.681
	Fuzzers		0.893	0.893	0.708
	Worms		0.000	0.052	0.000
	Backdoor		0.006	0.031	0.000
	Analysis		0.103	0.030	0.027
Result	W. Avg.	Recall	0.891	0.974	0.971
		Precision	—	0.981	—
		F-measure	—	0.976	—
			1.112	1.324	1.190
			0.119	0.132	0.090
			0.426	0.399	0.434
			0.290	0.297	0.287