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

Table 5

The performance of the RF model.

Feature extraction method	Matrix size	Testing performance				Cross-validation performance
		Accuracy	Precision	Recall	F1-score	Accuracy	Precision	Recall	F1-score
Unigram	1000	83.36	84.69	83.36	82.73	86.43 ± 0.48	87.41 ± 0.46	86.37 ± 0.49	86.02 ± 0.57
	3000	84.71	85.8	84.71	84.06	89.56 ± 0.34	90.05 ± 0.46	89.62 ± 0.34	89.3 ± 0.48
Bigram	1000	83.05	84.41	83.05	82.38	85.79 ± 0.51	86.87 ± 0.52	85.81 ± 0.49	85.4 ± 0.61
	3000	84.7	85.81	84.7	84.09	89.48 ± 0.36	89.79 ± 0.45	89.44 ± 0.35	89.12 ± 0.45
Trigram	1000	83.11	84.49	83.11	82.45	85.81 ± 0.46	86.93 ± 0.49	85.76 ± 0.41	85.4 ± 0.44
	3000	84.67	85.82	84.67	84.04	89.39 ± 0.44	89.9 ± 0.35	89.48 ± 0.39	89.2 ± 0.33
Four-gram	1000	83.07	84.49	83.07	82.46	85.29 ± 0.51	86.62 ± 0.57	85.34 ± 0.57	84.96 ± 0.56
	3000	84.61	85.82	84.61	84	89.41 ± 0.44	89.85 ± 0.41	89.4 ± 0.39	89.11 ± 0.41