Research Article

Two-Stage Bagging Pruning for Reducing the Ensemble Size and Improving the Classification Performance

Table 7

Relative improvement of the proposed bagging pruning methods with respect to the traditional bagging using GNB as the base classifier. The values in parentheses represent the optimized parameters ta for AP, td for DP, and ta, td for “AP+DP” and “DP+AP” with the best classification performance achieved by the corresponding pruning method.

DatasetAP“AP+DP”DP“DP+AP”

Aba0.8391.549,10.1970.192,7
Adult1.3592.279,10.0522.069,1
Aus4.9484.948,100.0090.009,8
Bcw0.0084.509,10.0042.339,1
Bld14.11916.039,12.57514.759,1
Cmc3.9494.469,11.0461.318,2
Col3.44914.179,10.87246.788,1
Cre5.2897.589,50.7327.589,6
Der26.79958.579,12.4112.415,1
Ger6.4386.437,80.7197.148,9
Gla19.38923.269,82.3272.325,7
Hea6.66911.168,26.66211.165,2
Hep21.87924.999,10.00229.699,1
Ion5.4195.418,25.4188.069,4
kr-vs-kp6.2186.218,60.5122.689,1
Mam3.4093.926,60.9843.406,6
Pid1.5994.769,40.5393.179,5
Spe2.6095.249,11.30311.849,1
Tel0.4890.849,30.1170.809,2
Veh3.6374.069,20.00100.657,6
Vot4.1794.179,40.0014.178,1
Vow5.4166.039,90.0090.000,9
Yea32.43932.849,82.35120.733,1
Spambase1.8094.269,10.1133.939,3
Tictacto3.4693.469,100.0070.363,8
Wdbc0.0058.626,12.7625.699,1
Wpbc8.45921.149,10.006,1033.808,1
Spect12.61920.719,34.502,109.009,3