Table 1.

Simulation results for network selection.

Network estimation accuracy
	TPR	FPR	F1	MCC	AUC	TPR	FPR	F1	MCC	AUC
	Random					Hub
mSSL-DPE	0.000	0.015	0.000	−0.019	0.499	0.001	0.024	0.001	−0.021	0.450
SpiecEasi	0.013	0.009	0.018	0.005	0.563	0.011	0.009	0.015	0.003	0.528
mLDM	0.277	0.181	0.067	0.039	0.560	0.440	0.248	0.080	0.069	0.602
SINC (B = 0)	0.276	0.225	0.056	0.020	0.542	0.253	0.241	0.038	0.004	0.507
SINC (τ = 1)	0.420	0.093	0.167	0.169	0.750	0.175	0.096	0.059	0.037	0.613
SINC (τ learned)	0.598	0.091	0.237	0.263	0.838	0.294	0.094	0.098	0.094	0.689
	Cluster					Band
mSSL-DPE	0.005	0.0181	0.008	−0.0230	0.446	0.003	0.022	0.004	−0.032	0.468
SpiecEasi	0.013	0.010	0.0182	0.005	0.563	0.012	0.009	0.020	0.007	0.544
mLDM	0.232	0.155	0.126	0.051	0.544	0.440	0.248	0.080	0.069	0.602
SINC (B = 0)	0.272	0.229	0.110	0.024	0.534	0.294	0.242	0.114	0.028	0.533
SINC (τ = 1)	0.294	0.084	0.223	0.169	0.678	0.311	0.088	0.230	0.175	0.685
SINC (τ learned)	0.411	0.080	0.306	0.261	0.741	0.446	0.089	0.312	0.269	0.737

NOTE: mSSL-DPE refers to the method of Deshpande, Ročková, and George (2019), SpiecEasi to the method of Kurtz et al. (2015), mLDM to Yang, Chen, and Chen (2017), SINC (B = 0) to the modified version of the proposed model with the covariate estimates fixed, and SINC to the proposed model. Random, Hub, Cluster, and Band refer to the underlying shape of the network, as illustrated in Figure 1. Bold values reflect top performing methods.