Human-in-the-loop review board for evaluating candidate causal models against baseline performance metrics.
| Evaluation Metric / Attribute | Baseline Model (v1.0) | Candidate Model (v2.0 - Certified) | Status / Delta |
|---|---|---|---|
| Out-of-Sample MAE (2023 Holdout) | 9.352 | 7.829 | Improved (-1.523) |
| Holdout Predictive R² Score | 0.412 | 0.828 | Superior (+0.416) |
| 50-Seed Monte Carlo Stability Range | [0.450, 0.710] | [0.5716, 0.6703] | High Stability |
| Covariate Data Drift Detected | Moderate Risk | Low Risk | Passed |
| Placebo Test Falsification | Passed (p < 0.01) | Passed (p < 0.001) | Certified |
Authorizing this change will instantly bind backend inference API calls to the certified candidate model pipeline.