phishGILLNET—phishing detection methodology using probabilistic latent semantic analysis, AdaBoost, and co-training

Table 7 phishGILLNET2--binary (phish versus not phish) classification performance

Topics	Weak learner for boosting	TPR	FPR	Precision	Recall	F-measure	ROC Area	Time (s)
50	C4.5	0.985	0.055	0.985	0.985	0.985	0.966	0.79
50	RIPPER	0.989	0.051	0.989	0.989	0.989	0.968	4.17
50	Random forest	0.993	0.053	0.993	0.993	0.993	0.999	1.31
50	SVM	0.939	0.355	0.935	0.939	0.937	0.792	12.67
50	Logistic	0.938	0.421	0.932	0.938	0.933	0.957	1.0
100	C4.5	0.995	0.02	0.995	0.995	0.995	0.987	1.58
100	RIPPER	0.997	0.012	0.997	0.997	0.997	0.993	6.82
100	Random forest	0.994	0.052	0.994	0.994	0.994	0.999	2.32
100	SVM	0.992	0.069	0.992	0.992	0.992	0.961	10.55
100	Logistic	0.995	0.023	0.995	0.995	0.995	0.994	2.17
200	C4.5	0.996	0.019	0.996	0.996	0.996	0.991	2.51
200	RIPPER	0.994	0.024	0.994	0.994	0.994	0.987	7.85
200	Random forest	0.995	0.037	0.995	0.995	0.995	0.999	2.87
200	SVM	0.988	0.098	0.988	0.988	0.988	0.945	10.78
200	Logistic	0.997	0.018	0.997	0.997	0.997	0.997	4.11