Table 2.

Test metrics of the 3 algorithms for all 3 classification tasks as well as average model performance at 500 features for each classification task.

Test metrics and performance	Logistic regression				Bernoulli naïve Bayes				Random forest
	Acca	F	Precb	Recc	Acc	F	Prec	Rec		Acc	F	Prec	Rec
Underage JUUL use	0.94	0.94	0.95	0.92	0.78	0.71	0.99	0.57		0.99	0.99	0.99	0.99
Positive sentiment	0.72	0.69	0.82	0.69	0.69	0.63	0.83	0.53		0.82	0.82	0.80	0.75
Negative sentiment	0.78	0.77	0.85	0.73	0.72	0.66	0.98	0.50		0.91	0.91	0.90	0.94
Average model performance	0.81	0.80	0.87	0.78	0.73	0.67	0.93	0.53		0.91	0.91	0.90	0.89

^aAcc: accuracy

^bPrec: precision

^cRec: recall