Table 5.

Overview of the results for FCS-001 and all courses combined. All experiments are performed using SVM and random oversampling of the minority class. The metrics are assessed by using 10-fold cross validation. The best result per metric for a combination of course, pattern type and support is shown in bold.

Dataset	Patterns	Support	w	#patterns	Acc	AUC	Lift
FCS-001	Abstract	0.4	1	70	0.59	0.52	1.29
FCS-001	Abstract	0.4	2	117	0.66	0.63	1.93
FCS-001	Abstract	0.4	4	115	0.65	0.58	1.72
FCS-001	Abstract	0.6	1	46	0.59	0.51	1.72
FCS-001	Abstract	0.6	2	49	0.64	0.54	1.29
FCS-001	Abstract	0.6	4	65	0.61	0.57	1.50
FCS-001	Abstract	0.8	1	20	0.50	0.38	0.21
FCS-001	Abstract	0.8	2	49	0.60	0.51	1.50
FCS-001	Abstract	0.8	4	29	0.67	0.56	1.72
FCS-001	Course-specific	0.4	1	818	0.66	0.57	1.50
FCS-001	Course-specific	0.4	2	533	0.65	0.52	1.29
FCS-001	Course-specific	0.4	4	446	0.67	0.55	1.07
FCS-001	Course-specific	0.6	1	617	0.66	0.57	2.15
FCS-001	Course-specific	0.6	2	284	0.61	0.54	1.29
FCS-001	Course-specific	0.6	4	232	0.61	0.54	0.86
FCS-001	Course-specific	0.8	1	267	0.65	0.60	1.07
FCS-001	Course-specific	0.8	2	83	0.65	0.58	1.07
FCS-001	Course-specific	0.8	4	100	0.57	0.51	0.43
All courses	Abstract	0.4	1	67	0.64	0.64	1.76
All courses	Abstract	0.4	2	99	0.59	0.61	1.71
All courses	Abstract	0.4	4	135	0.58	0.59	1.61
All courses	Abstract	0.6	1	84	0.62	0.60	1.70
All courses	Abstract	0.6	2	76	0.57	0.60	1.54
All courses	Abstract	0.6	4	43	0.61	0.61	1.83
All courses	Abstract	0.8	1	36	0.65	0.60	1.21
All courses	Abstract	0.8	2	14	0.68	0.60	1.43
All courses	Abstract	0.8	4	10	0.52	0.52	1.00