Time Series Forecasting on ETTm2 10% train (test)
0.257MSEFiCoTS
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| FiCoTSInput sequence length=512, Predictive lengths={96, 192, 336, 720}, Method Category=LLM-based, Training data percentage=10%2025.11 | 0.257 | 0.317 | |
| TimeVLMInput sequence length=512, Predictive lengths={96, 192, 336, 720}, Method Category=LLM-based, Training data percentage=10%2025.11 | 0.262 | 0.323 | |
| Time-VLMForecasting Horizon (H)=Avg2026.05 | 0.263 | 0.323 | |
| SSDAForecasting Horizon (H)=Avg2026.05 | 0.27 | 0.32 | |
| Time-LLMForecasting Horizon (H)=Avg2026.05 | 0.277 | 0.323 | |
| TIME-LLMParameters (M)=6.46, Memory (MiB)=3907, Eff.*MSE=0.43, Eff.*MAE=0.402026.03 | 0.28 | 0.33 | |
| One-for-AllParameters (M)=0.55, Memory (MiB)=2.2, Eff.*MSE=4.37, Eff.*MAE=4.482026.03 | 0.29 | 0.33 | |
| VisionTSForecasting Horizon (H)=Avg2026.05 | 0.292 | 0.342 | |
| GPT4TSForecasting Horizon (H)=Avg2026.05 | 0.293 | 0.335 | |
| S2IP-LLMInput sequence length=512, Predictive lengths={96, 192, 336, 720}, Method Category=LLM-based, Training data percentage=10%2025.11 | 0.296 | 0.342 | |
| PatchTSTForecasting Horizon (H)=Avg2026.05 | 0.296 | 0.343 | |
| GPT4TSParameters (M)=11.7, Memory (MiB)=371.0, Eff.*MSE=0.21, Eff.*MAE=0.212026.03 | 0.3 | 0.34 | |
| TimeLLMInput sequence length=512, Predictive lengths={96, 192, 336, 720}, Method Category=LLM-based, Training data percentage=10%2025.11 | 0.307 | 0.348 | |
| GPT4TSInput sequence length=512, Predictive lengths={96, 192, 336, 720}, Method Category=LLM-based, Training data percentage=10%2025.11 | 0.308 | 0.346 | |
| PatchTSTInput sequence length=512, Predictive lengths={96, 192, 336, 720}, Method Category=Transformer-based, Training data percentage=10%2025.11 | 0.314 | 0.352 | |
| DLinearForecasting Horizon (H)=Avg2026.05 | 0.316 | 0.368 | |
| TESTParameters (M)=2.00, Memory (MiB)=701, Eff.*MSE=1.24, Eff.*MAE=1.262026.03 | 0.32 | 0.31 | |
| TimesNetParameters (M)=0.63, Memory (MiB)=2.9, Eff.*MSE=3.02, Eff.*MAE=3.462026.03 | 0.32 | 0.35 | |
| TimesNetForecasting Horizon (H)=Avg2026.05 | 0.32 | 0.353 | |
| TStationaryParameters (M)=2.02, Memory (MiB)=13.2, Eff.*MSE=0.88, Eff.*MAE=1.052026.03 | 0.33 | 0.37 | |
| TimesNetInput sequence length=512, Predictive lengths={96, 192, 336, 720}, Method Category=Non-Transformer-based, Training data percentage=10%2025.11 | 0.344 | 0.372 | |
| iTransformerInput sequence length=512, Predictive lengths={96, 192, 336, 720}, Method Category=Transformer-based, Training data percentage=10%2025.11 | 0.356 | 0.388 | |
| FEDformerInput sequence length=512, Predictive lengths={96, 192, 336, 720}, Method Category=Transformer-based, Training data percentage=10%2025.11 | 0.381 | 0.404 | |
| AutoformerInput sequence length=512, Predictive lengths={96, 192, 336, 720}, Method Category=Transformer-based, Training data percentage=10%2025.11 | 0.388 | 0.433 | |
| DLinearInput sequence length=512, Predictive lengths={96, 192, 336, 720}, Method Category=Non-Transformer-based, Training data percentage=10%2025.11 | 0.399 | 0.426 | |
| ETSformerParameters (M)=5.28, Memory (MiB)=31.4, Eff.*MSE=0.25, Eff.*MAE=0.312026.03 | 0.45 | 0.49 | |
| EFDformerParameters (M)=16.8, Memory (MiB)=87.8, Eff.*MSE=0.12, Eff.*MAE=0.122026.03 | 0.46 | 0.49 | |
| FEDformerForecasting Horizon (H)=Avg2026.05 | 0.463 | 0.488 | |
| ETSformerInput sequence length=512, Predictive lengths={96, 192, 336, 720}, Method Category=Transformer-based, Training data percentage=10%2025.11 | 0.534 | 0.547 | |
| AutoformerParameters (M)=10.53, Memory (MiB)=62.6, Eff.*MSE=0.13, Eff.*MAE=0.162026.03 | 1.34 | 0.93 | |
| InformerParameters (M)=11.33, Memory (MiB)=65.8, Eff.*MSE=0.04, Eff.*MAE=0.092026.03 | 3.37 | 1.44 | |
| InformerForecasting Horizon (H)=Avg2026.05 | 3.37 | 1.44 | |
| ReformerParameters (M)=5.80, Memory (MiB)=33.4, Eff.*MSE=0.08, Eff.*MAE=0.092026.03 | 3.98 | 1.59 |