Table 4. Usability evaluation of the top-three ranked algorithms based on the overall performance measure.

Legend: NP: number of parameters; GP: Generalizability measure, normalized between 0 (no generalizability) and 1 (complete generalizability); TIM: execution time in seconds.

Color code: For each dataset and parameter, red background means the worst value of the three methods, yellow means the intermediate value, and green means the best value out of the three listed.

	1st ranked			2nd ranked			3rd ranked
	NP	GP	TIM	NP	GP	TIM	NP	GP	TIM
DIC-C2DH-HeLa	FR-Ro-GE 0.828			KTH-SE (4) 0.629			IMCB-SG (1) 0.523
	4	0.912	4818	14	0.928	622	5	0.924	236
Fluo-C2DL-MSC	KTH-SE (1) 0.676			FR-Ro-GE 0.636			NOTT-UK 0.546
	17	0.893	79	4	0.893	2630	5	0.920	342
Fluo-C3DH-H157	KTH-SE (1) 0.938			HD-Har-GE 0.885			CUNI-CZ 0.870
	17	0.966	16156	10	0.882	14110	8	0.836	952
Fluo-C3DL-MDA231	KTH-SE (1) 0.757			LEID-NL 0.745			IMCB-SG (2) 0.659
	16	0.947	217	9	0.958	992	9	0.936	3506
Fluo-N2DH-GOWT1	KTH-SE (1) 0.951			LEID-NL 0.902			CUNI-CZ 0.902
	17	0.955	632	9	0.932	1333	8	0.950	479
Fluo-N2DL-HeLa	KTH-SE (1) 0.942			FR-Ro-GE 0.940			HD-Har-GE 0.901
	17	0.967	304	3	0.963	22878	10	0.966	609
Fluo-N3DH-CE	KTH-SE (1) 0.688			HD-Har-GE 0.601			KIT-GE 0.507
	17	0.895	13475	9	0.889	14518	10	0.872	4258
Fluo-N3DH-CHO	KTH-SE (1) 0.926			MU-CZ 0.912			HD-Har-GE 0.906
	17	0.954	202	8	0.936	223	10	0.923	1495
Fluo-N3DL-DRO	KTH-SE (2) 0.609			UP-PT 0.285			CUL-UK 0.220
	20	0.885	85272	8	0.916	13772	3	0.973	6902
PhC-C2DH-U373	FR-Ro-GE 0.951			FR-Be-GE 0.896			KTH-SE (3) 0.886
	5	0.965	11450	8	0.953	621	11	0.964	81
PhC-C2DL-PSC	HD-Hau-GE 0.804			KTH-SE (1) 0.772			UP-PT 0.735
	15	0.952	924	17	0.971	3481	11	0.959	8246
Fluo-N2DH-SIM+	FR-Ro-GE 0.878			KTH-SE (1) 0.874			PAST-FR 0.859
	3	0.979	20124	17	0.983	301	9	0.978	370
Fluo-N3DH-SIM+	KTH-SE (1) 0.848			LEID-NL 0.798			IMCB-SG (2) 0.714
	17	0.985	13115	9	0.973	66773	9	0.988	69549