Table 3.

Pearson correlation coefficient (PCC) and root-mean-square error (RMSE) for influenza prediction models using different word embedding techniques.

Prediction model	PCC (number of keywords)		RMSE (number of keywords)
	Unsorted	Sorted	Unsorted	Sorted
Word2Vec CBOWa	0.8784 (59)	0.8951 (22)	0.0095 (19)	0.0082 (22)
Word2Vec skip-gram	0.8755 (50)	0.8942 (8)	0.0089 (9)	0.0080 (8)
GloVe	0.8467 (14)	0.8783 (29)	0.0095 (14)	0.0090 (22)
FastText CBOW	0.8845 (42)	0.8986 (34)	0.0095 (11)	0.0090 (34)
FastText skip-gram	0.8676 (86)	0.8679 (10)	0.0095 (87)	0.0090 (10)
Mean	0.8705 (50)	0.8868 (21)	0.0094 (28)	0.0086 (19)

^aCBOW: continuous bag-of-words.