Electricity Load Forecasting on Australian Electricity Load 2019-2021 (Half-Hourly)
201.736RMSEGRAFT (Reddit)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| GRAFT (Reddit)External Source=Reddit, Forecasting scale=daily-scale, Numerical Backbone=STanHop2025.12 | 201.736 | 170.002 | 0.0741 | |
| GRAFT (News)External Source=News, Forecasting scale=daily-scale, Numerical Backbone=STanHop2025.12 | 222.573 | 184.32 | 0.0701 | |
| GRAFT (All)External Source=Reddit + News + Policy, Forecasting scale=daily-scale, Numerical Backbone=STanHop2025.12 | 254.803 | 209.04 | 0.0854 | |
| STanHop (NoExt)External Source=None, Forecasting scale=daily-scale2025.12 | 256.517 | 211.975 | 0.0781 | |
| LLM-based agent 4Model Category=LLM-based agent, Reasoning=News-prompted2025.12 | 280.39 | 180.96 | 0.0515 | |
| GRAFT (Policy)External Source=Policy, Forecasting scale=daily-scale, Numerical Backbone=STanHop2025.12 | 287.581 | 232.257 | 0.1058 | |
| PyraformerModel Category=Pure numerical time-series model2025.12 | 312.42 | 220.32 | 0.0687 | |
| LLM-based agent 2Model Category=LLM-based agent, Reasoning=News-prompted2025.12 | 336.41 | 206.08 | 0.0529 | |
| LLM-based agent 1Model Category=LLM-based agent, Reasoning=News-prompted2025.12 | 337.1 | 204.89 | 0.0527 | |
| PatchTSTModel Category=Pure numerical time-series model2025.12 | 365.41 | 234.46 | 0.0656 | |
| FEDformerModel Category=Pure numerical time-series model2025.12 | 366 | 238.77 | 0.0675 | |
| TimesNetModel Category=Pure numerical time-series model2025.12 | 366.64 | 237.49 | 0.0681 | |
| iTransformerModel Category=Pure numerical time-series model2025.12 | 367.79 | 233.58 | 0.0686 | |
| GPT4TSModel Category=LLM-based forecasting method2025.12 | 377.62 | 236.91 | 0.0661 | |
| FiLMModel Category=Pure numerical time-series model2025.12 | 392.3 | 254.05 | 0.0736 | |
| DLinearModel Category=Pure numerical time-series model2025.12 | 401.98 | 255.7 | 0.0729 | |
| InformerModel Category=Pure numerical time-series model2025.12 | 407.52 | 282.56 | 0.0894 | |
| LLM-based agent 3Model Category=LLM-based agent, Reasoning=News-prompted2025.12 | 407.86 | 250.75 | 0.0684 | |
| AutoformerModel Category=Pure numerical time-series model2025.12 | 501.78 | 349.43 | 0.1063 |