Table 1.

Performance for each model based on variations of text and hashtag use.

Performance measure		Decision tree	Random forest	Support vector machine	Study model
Text with hashtags, %
	Precision	95.05	96.00	96.86	94.81
	Recall	82.15	86.08	81.21	91.42
	F1 score	88.13	90.77	88.35	93.09
	Area under the curve	96.67	96.85	97.18	98.12
Hashtags only, %
	Precision	86.22	94.14	95.39	89.60
	Recall	86.50	87.13	84.24	88.89
	F1 score	86.36	90.50	89.47	89.24
	Area under the curve	95.95	95.23	95.43	94.32
Text without hashtags, %
	Precision	88.49	97.07	97.80	93.60
	Recall	93.08	91.31	89.32	98.31
	F1 score	90.73	94.11	93.37	95.90
	Area under the curve	95.56	94.85	93.49	99.12