Table 1.

Mean accuracies (%, for 10-fold cross-validation) of various approaches for heartbeat classification.

Feature groups and subgroups	Classification model
	k-NN			DFA		NB		SVM
	k = 1	k = 5	k = 10	Linear	Quadratic	Gaussian	Kernel	RBF	Linear
CommonD	98.5	97.7	91.9	75.6	77.1	77.7	81.2	81.6	79.6
LoopD	97.6	94.8	87.6	63.1	60.7	59.6	80.6	66.3	64.6
AreaD	99.0	98.6	93.6	81.7	75.9	75.8	80.7	86.9	83.9
AreaDa	98.9	96.7	90.6	72.0	72.8	72.7	78.4	75.8	75.0
AreaDr	98.3	96.1	90.3	62.6	65.8	66.1	80.1	81.9	78.1
MorphD	99.0	98.0	94.2	77.9	74.0	74.4	83.5	93.5	87.9
CommonR	98.4	97.3	93.5	77.1	62.0	62.0	79.4	88.8	82.3
AreaR	98.2	94.2	88.7	50.3	56.0	57.3	83.2	63.3	61.7
AreaRa	99.0	98.3	91.6	68.1	66.3	66.4	83.4	85.7	80.7
MorphR	71.7	77.0	76.8	47.2	57.3	57.3	70.8	46.5	46.3
SpectralD	94.8	93.2	90.0	62.6	45.1	45.4	82.8	72.7	69.4
S20	98.8	96.1	92.0	50.4	23.3	23.5	83.5	79.6	61.7
S30	97.1	93.0	89.1	58.8	37.6	37.9	84.2	74.2	70.5
S50	98.0	95.8	91.1	60.3	34.7	34.8	87.4	82.0	79.2
SpectralR	97.3	96.2	91.2	50.6	33.9	34.0	86.1	82.1	80.1