An Optimized Hybrid Deep Learning Model to Detect COVID-19 Misleading Information

Table 14

The performance of applying ML to dataset 3.

Models	Feature selection methods	Cross-validation performance				Test performance
		ACC	PRE	REC	F1	ACC	PRE	REC	F1
DT	Unigram	72.21	72.5	72.33	72.07	66.96	66.7	66.96	66.81
	Bi-gram	72.11	72.1	71.91	71.84	64.77	66.18	64.77	65.27
	Tri-gram	68.82	71.98	68.79	69.46	61.14	66.94	61.14	62.14
	Four-gram	67.67	64.84	67.59	61.23	52.98	66.1	52.98	52.49
KNN	Unigram	73.16	76.98	73.32	71.69	69.12	75.72	69.12	59.7
	Bi-gram	70.45	75.12	70.45	62.69	68.17	73.48	68.17	57.92
	Tri-gram	73.32	72.61	73.16	70.95	67.52	69.88	67.52	56.99
	Four-gram	67.44	70.66	67.44	68.1	35.8	67.11	35.8	21.7
LR	Unigram	81.71	81.57	81.71	81.06	74.7	75.14	74.7	71.71
	Bi-gram	77.39	77.87	77.39	75.33	68.75	69.24	68.75	68.96
	Tri-gram	77.28	77.81	77.28	75.12	68.86	69.19	68.86	69.01
	Four-gram	68.87	69.42	68.87	61.02	60.63	69.25	60.63	61.4
RF	Unigram	80.2	79.84	79.48	78.21	75.32	75.98	75.32	72.45
	Bi-gram	73.39	76.34	73.87	68.5	65.71	58.29	65.71	54.0
	Tri-gram	67.82	72.71	67.98	57.21	65.71	58.63	65.71	54.26
	Four-gram	67.41	72.33	67.46	56.2	65.5	58.5	65.5	54.0
SVM	Unigram	82.12	82.01	82.12	81.63	76.92	76.49	76.92	75.54
	Bi-gram	74.71	76.61	74.71	70.95	66.35	62.89	66.35	55.36
	Tri-gram	69.5	71.57	69.5	61.72	66.03	60.97	66.03	55.42
	Four-gram	69.1	71.13	69.1	60.95	65.22	55.61	65.22	53.75
NB	Unigram	79.97	79.89	79.97	78.92	70.62	73.8	70.62	63.58
	Bi-gram	80.25	80.27	80.25	80.2	72.26	71.11	72.26	71.09
	Tri-gram	69.21	71.72	69.21	60.91	66.11	61.87	66.11	54.06
	Four-gram	69.1	71.16	69.23	60.17	65.74	58.39	65.74	53.86