Comparative study between incremental and ensemble learning on data streams: Case study

Journal of Big Data

Table 4 The mean of the accuracy on data stream Hyperplane3

	Data-chunk size	500	1000	2000
Classifier number	Algorithm
10	SEA	77.08_±9.59	78.44_±11.07	77.42_±4.03
	AWE	88.04_±2.52	86.99_±2.97	85.13_±2.20
20	SEA	79.4_±18.41	78.58_±5.78	75.8_±4.98
	AWE	88.26_±2.61	87.17_±3.12	85.53_±2.04
30	SEA	79.26_±6.22	78.82_±8.45	75.44_±1.84
	AWE	88.75_±2.36	87.86_±2.70	86.44_±1.84

SEA get the optimal accuracy at a 500 data chunk size and 20 base classifiers, while AWE algorithm get optimal accuracy at a 500 data chunk size and 30 base classifiers.