Selective Regression on cycle (test)
26.7Conditional Large-Loss RateLocus-BART
Evaluation results
| Method | Links | |
|---|---|---|
| Locus-BARTAcceptance Rate=≈ 70%, Calibration Level (alpha)=10%, Score Function=loss-quantile score, Gamma correction=false2026.03 | 26.7 | |
| Locus-BART (γ)Acceptance Rate=≈ 70%, Calibration Level (alpha)=10%, Score Function=loss-quantile score, Gamma correction=true2026.03 | 27.1 | |
| Locus-BARTMethod category=Locus-Tuned, Base model=BART, Calibration level (alpha)=10%, Target acceptance rate=≈ 70%2026.03 | 27.5 | |
| Locus-BART (gamma-infl.)Method category=Locus-Tuned, Base model=BART, Inflation=gamma-infl., Calibration level (alpha)=10%, Target acceptance rate=≈ 70%2026.03 | 28 | |
| Locus-MCAcceptance Rate=≈ 70%, Calibration Level (alpha)=10%, Score Function=loss-quantile score, Gamma correction=false2026.03 | 28.4 | |
| Locus-MC DropoutMethod category=Locus-Tuned, Base model=MC Dropout, Calibration level (alpha)=10%, Target acceptance rate=≈ 70%2026.03 | 28.7 | |
| IFlagAcceptance Rate=≈ 70%2026.03 | 28.9 | |
| Locus-MC (γ)Acceptance Rate=≈ 70%, Calibration Level (alpha)=10%, Score Function=loss-quantile score, Gamma correction=true2026.03 | 28.9 | |
| IFlagMethod category=Baseline (no Locus), Base model=IFlag, Target acceptance rate=≈ 70%2026.03 | 28.9 | |
| Locus-MC Dropout (gamma-infl.)Method category=Locus-Tuned, Base model=MC Dropout, Inflation=gamma-infl., Calibration level (alpha)=10%, Target acceptance rate=≈ 70%2026.03 | 29.2 | |
| VARNetAcceptance Rate=≈ 70%2026.03 | 29.7 | |
| VARNetMethod category=Baseline (no Locus), Base model=VARNet, Target acceptance rate=≈ 70%2026.03 | 29.7 |