Table 1.

Performance change of stacking models with different training dataset sizes.

		Cohen's kappa			Relative Cohen's kappa
Percent of total training data	Number of training data	FC	HC	HC & HTL	FC	HC	HC & HTL
25.0%	185	0.546	0.556	0.642	0.851	0.866	1.000
37.5%	278	0.600	0.606	0.666	0.899	0.910	1.000
50.0%	370	0.655	0.679	0.731	0.897	0.929	1.000
62.5%	462	0.675	0.710	0.756	0.893	0.939	1.000
75.0%	554	0.707	0.744	0.788	0.897	0.945	1.000
87.5%	647	0.727	0.784	0.805	0.903	0.975	1.000
100.0%	739	0.727	0.796	0.809	0.899	0.984	1.000