Predicting Coronavirus Pandemic in Real-Time Using Machine Learning and Big Data Streaming System

Table 3

The performance of the KNN model.

Feature extraction method	Matrix size	Testing performance				Cross-validation performance
		Accuracy	Precision	Recall	F1-score	Accuracy	Precision	Recall	F1-score
Unigram	1000	62.15	69.94	62.15	58.99	65.75 ± 0.52	73.33 ± 0.65	65.75 ± 0.52	63.5 ± 0.6
	3000	63.77	70.11	63.77	59.36	68.36 ± 0.61	74.72 ± 0.54	68.36 ± 0.61	65.7 ± 0.76
Bigram	1000	62.97	70.96	62.97	59.5	66.09 ± 0.59	73.85 ± 0.89	66.09 ± 0.59	63.74 ± 0.76
	3000	64.49	71.02	64.49	59.96	69.13 ± 0.76	76.04 ± 0.56	69.13 ± 0.76	66.44 ± 0.97
Trigram	1000	63	70.72	63	59.57	66.08 ± 0.54	73.69 ± 0.71	66.08 ± 0.54	63.75 ± 0.65
	3000	64.54	70.69	64.54	60.07	69.07 ± 0.75	75.61 ± 0.66	69.07 ± 0.75	66.39 ± 0.96
Four-gram	1000	62.93	71.24	62.93	59.53	66.09 ± 0.63	73.76 ± 0.93	66.09 ± 0.63	63.75 ± 0.8
	3000	64.62	71.04	64.62	60.06	69.25 ± 0.82	76.16 ± 0.66	69.25 ± 0.82	66.56 ± 1.05