Probabilistic Forecasting on San Francisco wind dataset (test)
98.55PICPDeep AR
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Deep ARTarget coverage (1 - alpha)=0.952024.12 | 98.55 | 6.2023 | 6.29 | |
| LSTM + QD+Target coverage (1 - alpha)=0.95, Backbone=LSTM, Loss=QD+2024.12 | 97.34 | 6.043 | 6.21 | |
| GRU + QD+Target coverage (1 - alpha)=0.95, Backbone=GRU, Loss=QD+2024.12 | 97.24 | 6.309 | 6.49 | |
| LSTM + QuantileTarget coverage (1 - alpha)=0.95, Backbone=LSTM, Loss=Quantile2024.12 | 95.94 | 5.407 | 5.64 | |
| LSTM + TubeTarget coverage (1 - alpha)=0.95, Backbone=LSTM, Loss=Tube2024.12 | 95.61 | 4.627 | 4.84 | |
| TCN + TubeTarget coverage (1 - alpha)=0.95, Backbone=TCN, Loss=Tube2024.12 | 95.43 | 4.56 | 4.78 | |
| GRU + QuantileTarget coverage (1 - alpha)=0.95, Backbone=GRU, Loss=Quantile2024.12 | 95.1 | 4.955 | 5.21 | |
| GRU + TubeTarget coverage (1 - alpha)=0.95, Backbone=GRU, Loss=Tube2024.12 | 95.07 | 4.857 | 5.11 | |
| TCN + QD+Target coverage (1 - alpha)=0.95, Backbone=TCN, Loss=QD+2024.12 | 95.05 | 5.49 | 5.78 | |
| MDNTarget coverage (1 - alpha)=0.952024.12 | 94.9 | 4.62 | 4.87 | |
| TCN + QuantileTarget coverage (1 - alpha)=0.95, Backbone=TCN, Loss=Quantile2024.12 | 94.28 | 4.908 | 5.21 | |
| TimeGPTTarget coverage (1 - alpha)=0.952024.12 | 93.57 | 11.235 | 12 |