Prevalence Estimation on CAP Law & Crime Spain media Claude Opus 4.6 (test)
-12.2Prevalence Estimation Bias (pp)Importance-weighted estimation
Evaluation Results
| Method | Links | |
|---|---|---|
| Importance-weighted estimationInput=Target-specific density ratio2026.04 | -12.2 | |
| MCGrad (scores)Input=Probability scores2026.04 | -4.5 | |
| Isotonic regressionInput=Probability scores2026.04 | -2.5 | |
| MCGrad (binary)Input=Binary labels, Initialization=Base-rate2026.04 | -1.9 | |
| Rogan-GladenInput=Binary labels2026.04 | 3.1 | |
| Classify & CountInput=Fraction of Yes labels2026.04 | 3.6 | |
| Saerens-Latinne-DecaesteckerInput=Probability scores2026.04 | 21.6 |