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

Table 5

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

CRC		Baseline	Tml	RUS	ENN	HDUS
KNN	Sensitivity (%)	29.41	37.25	62.75	62.75	80.35
	Specificity (%)	85.38	75.38	60	60.31	50.23
	Precision (%)	44.12	37.25	38.1	38.51	33.91
	F1_m (%)	35.29	37.25	47.41	47.73	47.69
	Bacc (%)	57.4	56.32	61.38	61.53	65.29
SVM	Sensitivity (%)	5.88	29.41	62.75	47.06	76.40
	Specificity (%)	92.31	82.31	54.62	70.77	55.85
	Precision (%)	23.08	39.47	35.16	38.71	35.40
	F1_m (%)	9.37	33.71	45.07	42.48	48.38
	Bacc (%)	49.1	55.86	58.68	58.91	66.13
DT	Sensitivity (%)	35.29	45.1	62.75	66.67	81.00
	Specificity (%)	68.46	59.23	46.92	43.85	56.91
	Precision (%)	30.51	30.26	31.68	31.78	39.90
	F1_m (%)	32.73	36.22	42.1	43.04	53.46
	Bacc (%)	51.88	52.16	54.83	55.26	68.96
AVG	Sensitivity (%)	23.53	37.25	62.75	58.83	79.25
	Specificity (%)	82.05	72.31	53.85	58.31	54.33
	Precision (%)	32.57	35.66	34.98	36.33	36.40
	F1_m (%)	27.32	36.44	44.92	44.92	49.85
	Bacc (%)	52.79	54.78	58.3	58.57	66.79