Table 3.

Predictive performances of the ensemble models based on the two candidate models using three labels (0: live, 1: early apoptotic, 2: late apoptotic).

Candidate Model	m00	m01	m02	m10	m11	m12	m20	m21	m22	Live precision	Live recall	Live f value	Early precision	Early recall	Early f value	Late precision	Late recall	Late f value	Accuracy
RF 76	1019	57	39	35	89	23	42	38	656	0.930	0.914	0.922	0.484	0.605	0.538	0.914	0.891	0.902	0.883
MLP 7-2	1016	62	37	30	93	24	44	32	660	0.932	0.911	0.922	0.497	0.633	0.557	0.915	0.897	0.906	0.885
hard_vote	1056	35	24	39	87	21	55	49	632	0.918	0.947	0.932	0.509	0.592	0.547	0.934	0.859	0.895	0.888
soft_vote	1026	50	39	34	90	23	42	34	660	0.931	0.920	0.926	0.517	0.612	0.561	0.914	0.897	0.905	0.889