Causal discovery on Sangiovese real-world fixed sample size
0.765RNBICA-LiNGAM
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ICA-LiNGAM2026.06 | 0.765 | 39.3 | |
| GES2026.06 | 0.569 | 45.1 | |
| DAGBag2026.06 | 0.539 | 43.1 | |
| LLM-BFSbackbone=GPT-5.52026.06 | 0.539 | 42.1 | |
| FCI2026.06 | 0.48 | 53.2 | |
| LLM-Greedybackbone=GPT-5.52026.06 | 0.471 | 59.1 | |
| PC2026.06 | 0.441 | 55.7 | |
| GRaSP2026.06 | 0.431 | 59.7 | |
| Uniform CEAmode=uniform weights2026.06 | 0.412 | 61.2 | |
| CEAbackbone=GPT-5.52026.06 | 0.373 | 65.2 | |
| BOSS2026.06 | 0.343 | 68.3 |