Table 3.

Performance of classification models.

Models		Stigma / nonstigma	Unpredictable / other subcategories	Dangerous / other subcategories	Depression / schizophrenia	Unpredictable (depression) / unpredictable (schizophrenia)	Glorified (depression) / glorified (schizophrenia)
Support vector machine
	Precision	.70	.72	.76	.80	.93	.65
	Recall	.70	.72	.76	.79	.92	.65
	F measure	.70	.72	.76	.79	.92	.65
Naïve Bayes
	Precision	.65	.67	.69	.67	.88	.67
	Recall	.60	.64	.66	.65	.88	.65
	F measure	.57	.62	.64	.64	.88	.64
Multilayer perceptron neural network
	Precision	.67	.69	.71	.75	.88	.71
	Recall	.67	.69	.71	.75	.88	.71
	F measure	.67	.69	.71	.75	.88	.71
Logistic model trees
	Precision	.89	.70	.75	.78	.91	.65
	Recall	.89	.70	.75	.77	.91	.65
	F measure	.89	.70	.75	.77	.91	.65