Causal structure learning on Synthetic nD=10000, d=2, dmax=10, 200 nodes
1.14Number of CI TestsMB-by-MB+
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| MB-by-MB+CI Test Type=d-separation2025.10 | 1.14 | 0.42 | 0 | 0.019 | |
| LOADCI Test Type=d-separation2025.10 | 1.49 | 0.55 | 100 | 0.004 | |
| LDP+CI Test Type=d-separation2025.10 | 1.51 | 2.44 | 0 | 0.007 | |
| SNAP(∞)CI Test Type=d-separation2025.10 | 20.02 | 14.42 | 100 | 0.004 | |
| MARVELCI Test Type=d-separation2025.10 | 21.11 | 13.23 | 100 | 0.004 | |
| LDECC+CI Test Type=d-separation2025.10 | 23 | 12.87 | 0 | 0.019 | |
| PCCI Test Type=d-separation2025.10 | 28.61 | 19.67 | 100 | 0.004 | |
| MB-by-MB+CI Test Type=G22025.10 | 1,710 | 4.16 | 0 | 0.026 | |
| LDP+CI Test Type=G22025.10 | 1,930 | 7.85 | 0 | 0.025 | |
| LOADCI Test Type=G22025.10 | 2,000 | 4.65 | 23 | 0.024 | |
| MARVELCI Test Type=G22025.10 | 19,900 | 1,068.25 | 0 | 0.055 | |
| SNAP(∞)CI Test Type=G22025.10 | 19,900 | 13.02 | 0 | 0.05 | |
| PCCI Test Type=G22025.10 | 22,770 | 13.84 | 31 | 0.017 |