Multivariate long-term forecasting on Traffic standard (test)
0.126MSEMulti-Scale Residual-Aware Representation Learning Pipeline
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Multi-Scale Residual-Aware Representation Learning PipelinePrediction Length (S)=96, Input Length=962026.06 | 0.126 | 0.085 | |
| TimeMixerAveraged over horizons={96, 192, 336, 720}2026.04 | 0.388 | 0.263 | |
| PatchTSTAveraged over horizons={96, 192, 336, 720}2026.04 | 0.391 | 0.264 | |
| TSMixerAveraged over horizons={96, 192, 336, 720}2026.04 | 0.408 | 0.284 | |
| iTransformerlook-back length L=96, prediction lengths S={96, 192, 336, 720}2024.02 | 0.428 | 0.282 | |
| ST-PTAveraged over horizons={96, 192, 336, 720}2026.04 | 0.446 | 0.289 | |
| TimeXerlook-back length L=96, prediction lengths S={96, 192, 336, 720}2024.02 | 0.466 | 0.287 | |
| PatchTSTlook-back length L=96, prediction lengths S={96, 192, 336, 720}2024.02 | 0.481 | 0.304 | |
| CrossformerAveraged over horizons={96, 192, 336, 720}2026.04 | 0.542 | 0.283 | |
| Crossformerlook-back length L=96, prediction lengths S={96, 192, 336, 720}2024.02 | 0.55 | 0.304 | |
| TimesNetlook-back length L=96, prediction lengths S={96, 192, 336, 720}2024.02 | 0.62 | 0.336 | |
| TimesNetAveraged over horizons={96, 192, 336, 720}2026.04 | 0.62 | 0.336 | |
| DLinearlook-back length L=96, prediction lengths S={96, 192, 336, 720}2024.02 | 0.625 | 0.383 | |
| RLinearlook-back length L=96, prediction lengths S={96, 192, 336, 720}2024.02 | 0.626 | 0.378 | |
| Autoformerlook-back length L=96, prediction lengths S={96, 192, 336, 720}2024.02 | 0.628 | 0.379 | |
| TiDElook-back length L=96, prediction lengths S={96, 192, 336, 720}2024.02 | 0.76 | 0.473 | |
| SCINetlook-back length L=96, prediction lengths S={96, 192, 336, 720}2024.02 | 0.804 | 0.509 |