Table 8:

Classification results when using various features.

		F1 Score	Precision	Recall	Specificity	log(C)	log(γ)
iOS	Volume-based features	0.60	0.61	0.59	0.62	−2	4
	Aggregate usage features	0.63	0.67	0.60	0.59	8	5
	Category-based usage features	0.65	0.62	0.68	0.61	−1	5
	Aggregate usage + volume-based features	0.69	0.62	0.78	0.62	−3	4
	Aggregate usage + category-based usage features	0.71	0.71	0.71	0.63	−2	4
	Aggregate usage + category-based usage + volume-based features	0.71	0.67	0.76	0.63	−2	4
Android	Volume-based features	0.56	0.50	0.64	0.45	3	1
	Aggregate usage features	0.62	0.7	0.57	0.51	0	1
	Category-based usage features	0.66	0.77	0.59	0.44	2	5
	Aggregate usage + volume-based features	0.71	0.71	0.71	0.69	5	2
	Aggregate usage + category-based usge features	0.80	0.75	0.86	0.77	8	−5
	Aggregate usage + category-based usage + volume-based features	0.80	0.75	0.86	0.77	8	−5