Table 5.

Optimal results of Models 4 and 5 on the validation set

Model name	Construction method	Optimal Vote difference	Number of optimal feature	Negative training n sample	Positive training n sample	Negative validation n sample	Positive validation n sample	Validation accuracy	Validation F1 score
Model4	Easy ensemble	15	45	175,109	10,714	43,261	2732	0.95	0.78
Model5	W-easy ensemble	9	45	175,109	10,714	43,261	2732	0.95	0.78