One-step-ahead forecasting on Hong Kong HFMD dataset (2023–2024, 90 weeks)
3.49MAEProphet
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Prophet2025.11 | 3.49 | 4.83 | 2.31 | 40 | |
| XGBoost2025.11 | 3.49 | 4.83 | 2.53 | 33.7 | |
| NSF-LLMBackbone=GPT-5.12025.11 | 3.49 | 4.815 | 2.08 | 87.9 | |
| ARIMA2025.11 | 3.51 | 4.83 | 2.54 | 32.6 | |
| LSTM2025.11 | 3.53 | 5.15 | 1.79 | 73.7 | |
| TimesFMPretrained=true2025.11 | 3.67 | 6.66 | 2.44 | 85.9 | |
| ChronosPretrained=true2025.11 | 3.79 | 5.23 | 1.71 | 56.8 | |
| MoiraiPretrained=true2025.11 | 3.95 | 5.73 | 1.86 | 83.2 | |
| NSF-LLMBackbone=Qwen3-235B2025.11 | 3.95 | 5.791 | 2.33 | 91.2 | |
| NSF-LLMBackbone=DeepSeek2025.11 | 4.32 | 5.87 | 2.4 | 90.1 |