Table 6.

Splitting the data.

	Protocol Split		Random Split		Oversampling
	Train	Test	Train	Test	Train	Test
number of N	45,866	44,259	63,150	26,975	63,050	27,075
number of S	944	1837	1973	808	63,225	26900
number of V	3788	3221	4845	2164	63042	27,083
number of F	415	388	536	267	63,076	27049
number of Q	8	7	9	6	63,044	27,081
Total	51,021	49,712	70,513	30,220	315,437	135,188