HDEC: A Heterogeneous Dynamic Ensemble Classifier for Binary Datasets

Table 4

The best accuracy of the ensemble approaches.

Dataset	Best-bagging		Best-AdaBoost		Best-stacking		CSS [36]	HDEC
	Type	Acc	Type	Acc	Type (meta)	Acc
AUS	DT	86.52 ± 0.97	DS	86.67 ± 0.82	BN	86.38 ± 1.12	—	87.12 ± 1.09
BRC	BN	73.95 ± 1.30	DS	71.61 ± 1.02	BN	72.73 ± 1.31	74.83 ± 0.70	76.47 ± 0.52
COL	RT	85.33 ± 0.36	DS	82.36 ± 1.46	BN	84.96 ± 0.26	86.74 ± 0.84	87.23 ± 0.60
DBT	BN	77.36 ± 0.50	BN	76.86 ± 0.31	NB	76.26 ± 0.25	77.86 ± 0.56	78.29 ± 0.31
CLS	KNN	100	KNN	100	All– (K, DsT)	100	—	100
HRT	NB	84.44 ± 0.21	SVM	84.44 ± 0.21	RT	84.44 ± 0.21	—	84.63 ± 0.41
HPT	NB	83.94 ± 0.73	NB	84.00 ± 0.69	BN	83.01 ± 0.30	86.52 ± 1.12	87.47 ± 0.97
ION	RT	93.16 ± 0.28	NB	93.15 ± 0.27	BN	93.45 ± 0.44	—	93.36 ± 0.21
LVD	DT	70.03 ± 1.63	DT	68.12 ± 1.15	NB	66.09 ± 1.32	71.51 ± 1.27	74.67 ± 1.02
SON	K	85.10 ± 0.27	KNN	86.54 ± 0.22	RT	88.94 ± 0.37	—	90.12 ± 0.44
GDE	DT	83.11 ± 1.24	DT	83.11 ± 1.13	SVM	83.75 ± 1.18	—	84.21 ± 0.97
VOT	DT	96.30 ± 0.36	NB	95.95 ± 0.33	DS	96.07 ± 0.59	96.90 ± 0.31	96.09 ± 0.21