Table 1.
Lasso performance with optimal λ1 value determined by BIC and mBIC criteria
| Nodes | Sample size | Criteria | λ1 | F-score | Recovered edges |
|---|---|---|---|---|---|
| 40 | 10 | BIC | 0.51 (0.07) | 0.16 (0.05) | 10 (3) |
| mBIC | 0.41 (0.07) | 0.19 (0.05) | 14 (4) | ||
| 100 | BIC | 0.21 (0.02) | 0.69 (0.04) | 58 (4) | |
| mBIC | 0.21 (0.01) | 0.69 (0.04) | 58 (4) | ||
| 200 | BIC | 0.14 (0.01) | 0.80 (0.03) | 72 (3) | |
| mBIC | 0.14 (0.01) | 0.80 (0.03) | 72 (3) | ||
| 300 | 10 | BIC | 0.49 (0.07) | 0.003 (0.001) | 2 (1) |
| mBIC | 0.39 (0.02) | 0.04 (0.01) | 33 (5) | ||
| 100 | BIC | 0.46 (0.04) | 0.13 (0.02) | 90 (5) | |
| mBIC | 0.25 (0.02) | 0.30 (0.01) | 234 (9) | ||
| 200 | BIC | 0.24 (0.02) | 0.39 (0.07) | 237 (9) | |
| mBIC | 0.20 (0.01) | 0.52 (0.02) | 391 (9) |
Note: The standard errors in parenthesis are calculated based on 100 simulated datasets. Recovered edges represent the total number of true edges recovered by the method.