[Retracted] The Use of Hellinger Distance Undersampling Model to Improve the Classification of Disease Class in Imbalanced Medical Datasets

Table 8

Evaluation results for BC dataset using classifiers (KNN, SVM, and DT) and models (baseline, Tml, RUS, ENN, and HDUS).

BC		Baseline	Tml	RUS	ENN	HDUS
KNN	Sensitivity (%)	33.33	44.44	57.11	50	61.11
	Specificity (%)	84.91	70.36	65.81	70.36	73.58
	Precision (%)	42.86	40	40.74	42.86	44.00
	F1_m (%)	37.5	42.1	47.56	46.16	51.16
	Bacc (%)	59.12	57.4	61.46	60.18	67.35
SVM	Sensitivity (%)	22.22	38.89	66.67	44.44	66.67
	Specificity (%)	94.34	88.68	64.15	83.02	69.81
	Precision (%)	57.14	53.85	38.71	47.06	42.86
	F1_m (%)	32	45.16	48.98	45.71	52.17
	Bacc (%)	58.28	63.78	65.41	63.73	68.24
DT	Sensitivity (%)	38.89	38.89	44.44	44.44	77.78
	Specificity (%)	66.04	64.15	62.26	69.81	66.04
	Precision (%)	28	26.92	28.57	33.33	43.75
	F1_m (%)	32.56	31.82	34.78	38.09	56.00
	Bacc (%)	52.46	51.52	53.35	57.13	71.91
AVG	Sensitivity (%)	31.48	40.74	56.07	46.29	68.52
	Specificity (%)	81.76	74.4	64.07	74.4	69.81
	Precision (%)	42.67	40.26	36.01	41.08	43.54
	F1_m (%)	36.23	40.5	43.85	43.53	53.11
	Bacc (%)	56.62	57.57	60.07	60.35	69.17