A Dynamic Ensemble Framework for Mining Textual Streams with Class Imbalance

Table 3

Kappa statistic of different algorithms on all streams.

	Spam	Spam-Enron	Spam1	Spam2	Reuters-Spam
CFIM	70.54	62.66	62.31	32.81	3.89
AUE-RT	55.07	43.67	50.92	31.86	3.21
AUE-RF	65.64	55.96	62.91	29.85	0.38
AUE-SVM	62.41	41.77	51.37	23.12	0.00
AWE-RT	47.05	42.13	32.43	4.90	3.24
AWE-RF	58.00	55.45	42.36	3.94	1.88
AWE-SVM	54.77	35.46	33.21	3.03	0.00
LB-RT	27.50	49.10	39.36	17.70	0.01
LB-RF	24.99	61.44	60.54	0.79	0.00
LB-SVM	45.29	46.78	16.30	0.00	0.00
OZA-RT	14.70	56.09	43.89	6.09	0.00
OZA-RF	31.44	58.76	54.45	5.92	0.00
OZA-SVM	36.42	53.38	47.32	6.29	0.00