Table 2.

Comparison of regression performance of node features and edge features by different methods

Method		RMSE (mean ± SD)		P-value	CC (mean ± SD)		P-value
		Train	Test		Train	Test
SM	Node	0.6837 ± 0.0101	0.6917 ± 0.0201	< 1e-3	0.2028 ± 0.0261	0.1787 ± 0.1050	<1e-3
	Edge	0.6950 ± 0.0061	0.7142 ± 0.0554	<1e-3	0.0785 ± 0.0201	0.0068 ± 0.1120	<1e-3
MM	Node	0.4528 ± 0.0098	0.5602 ± 0.0285	<1e-3	0.7410 ± 0.0239	0.5696 ± 0.1016	<1e-3
	Edge	0.5856 ± 0.0066	0.6635 ± 0.0254	<1e-3	0.5057 ± 0.0191	0.2380 ± 0.1170	<1e-3
DGMM	Node	0.4572 ± 0.0032	0.5526 ± 0.0032	0.0426	0.7439 ± 0.0012	0.5799 ± 0.0043	0.0396
	Edge	0.5891 ± 0.0024	0.6621 ± 0.0015	<1e-3	0.5131 ± 0.0026	0.2333 ± 0.0044	<1e-3
DAMM	Node	0.4648 ± 0.0132	0.5498 ± 0.0033	–	0.7571 ± 0.0077	0.5854 ± 0.0087	–
	Edge	0.5869 ± 0.0029	0.6577 ± 0.0019	–	0.5227 ± 0.0070	0.2446 ± 0.0013	–

Note: The best results are highlighted in bold.