A Dynamic Ensemble Framework for Mining Textual Streams with Class Imbalance

Table 2

Average accuracy of different algorithms on all streams.

	Spam	Spam-Enron	Spam1	Spam2	Reuters-Spam
CFIM	88.68	88.37	91.78	93.67	88.70
AUE-RT	81.73	81.37	82.24	88.26	80.40
AUE-RF	86.59	87.28	89.75	89.68	85.72
AUE-SVM	85.33	85.86	85.94	89.33	85.74
AWE-RT	77.24	75.71	70.69	46.74	78.31
AWE-RF	82.45	78.91	75.54	57.84	84.94
AWE-SVM	81.11	72.50	71.38	54.52	85.27
LB-RT	64.68	74.65	73.42	87.92	85.26
LB-RF	63.47	79.30	84.27	88.70	85.27
LB-SVM	76.09	72.68	50.69	88.66	85.27
OZA-RT	54.34	80.74	75.19	74.59	85.27
OZA-RF	67.80	81.56	79.39	74.07	85.27
OZA-SVM	70.88	72.50	78.28	74.33	85.27