Probabilistic Load Forecasting on PJM
0.0289CRPSBayesian Transformer
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Bayesian TransformerForecast Horizon (H)=24h2026.03 | 0.0289 | — | — | 1,573 | 2,143 | 90.4 | 4,960 | 0.0598 | |
| BT (Ours)Horizon (H)=24h2026.03 | 0.0289 | — | — | — | — | 90.4 | — | — | |
| Deep Ensemble (5×)Forecast Horizon (H)=24h, Ensemble size=5×2026.03 | 0.0312 | — | — | 1,589 | 2,176 | 89.2 | 5,480 | 0.0643 | |
| Deep EnsembleHorizon (H)=24h2026.03 | 0.0312 | — | — | — | — | 89.2 | — | — | |
| CQR (PatchTST)Forecast Horizon (H)=24h2026.03 | 0.0328 | — | — | 1,634 | 2,257 | 90.1 | 6,120 | 0.0671 | |
| CQRHorizon (H)=24h2026.03 | 0.0328 | — | — | — | — | 90.1 | — | — | |
| Quantile LSTMForecast Horizon (H)=24h2026.03 | 0.0351 | — | — | 1,698 | 2,344 | 87.6 | 5,940 | 0.0714 | |
| Quant. LSTMHorizon (H)=24h2026.03 | 0.0351 | — | — | — | — | 87.6 | — | — | |
| Standard TransformerForecast Horizon (H)=24h2026.03 | 0.0387 | — | — | 1,621 | 2,218 | 83.1 | 6,870 | 0.0823 | |
| Std. TransformerHorizon (H)=24h2026.03 | 0.0387 | — | — | — | — | 83.1 | — | — | |
| Det. LSTMForecast Horizon (H)=24h2026.03 | 0.0412 | — | — | 1,842 | 2,531 | 81.3 | 7,210 | 0.0891 | |
| Det. LSTMHorizon (H)=24h2026.03 | 0.0412 | — | — | — | — | 81.3 | — | — | |
| Multi-APLF2025.12 | 0.26 | 0.13 | 0.13 | — | — | — | — | — | |
| APLF2025.12 | 0.35 | 0.15 | 0.17 | — | — | — | — | — | |
| MTGP2025.12 | 0.42 | 0.21 | 0.21 | — | — | — | — | — |