A Lightweight Data Preprocessing Strategy with Fast Contradiction Analysis for Incremental Classifier Learning

Table 2

Performance comparison of algorithms under incremental learning mode.

Algorithm	Incremental learning
	No preprocessing			PP. with respect to local reference			PP. with respect to global reference
	Accuracy	Time	Gain ratio	Accuracy	Time	Gain ratio	Accuracy	Time	Gain ratio
J48	56.7105	12.0000	4.7259	71.6697	5.3333	13.4381	69.2177	12.5000	5.5374
NB	55.7100	2.6667	20.8913	68.5534	4.6667	14.6900	84.5540	2.0000	42.2770
NN	53.7179	834.0000	0.0644	66.2704	521.3333	0.1271	73.4672	984.0000	0.0747
SVM	55.8189	844.0000	0.0661	69.6478	238.6667	0.2918	76.5615	1040.0000	0.0736
Kstar	55.9808	3485.3333	0.0161	74.2809	1300.6667	0.0571	78.4271	3262.5000	0.0240
IBK	52.8808	63.3333	0.8350	69.1130	24.0000	2.8797	66.2719	61.5000	1.0776
LWL	59.5147	2562.6667	0.0232	73.4821	911.3333	0.0806	83.6198	4102.0000	0.0204
DT	56.9724	9.3333	6.1042	71.4792	4.6667	15.3170	69.0701	9.0000	7.6745