Table 2.

AUROC Area Under the Receiver Operator Curve (AUROC)

p = 100	‘un-shrunk’	‘shrunk’	gLASSO**
n = 10	0.732364	0.727612	0.6232
n = 20	0.819128	0.814088	0.7409
n = 30	0.862744*	0.8546	0.811604
n = 40	0.884656*	0.87488	0.83344
n = 50	0.899008*	0.889684	0.881036
n = 60	0.90848*	0.89878	0.888068
n = 70	0.914616*	0.904952	0.890632
n = 80	0.920464*	0.90786	0.906792
n = 90	0.929408*	0.917688	0.91554
n = 100	0.929864*	0.91908	0.915912

The table provides the average AUROC scores across 25 data sets for a network of p = 100 nodes with 3% of true positives edges and varying sample size n

*Two-sides t-test between the ‘un-shrunk’ and ‘shrunk’ methods statistically significant at 0.05

**STAR model selection is used from the R package ‘huge’[15, 19]