Model Training and Parameter Count on Traffic Time Series Forecasting
9,330,000ParametersCI-TSMixer (G,H,CC)
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| CI-TSMixer (G,H,CC)Gated Attention=true, Hierarchy Reconciliation=true, Cross-channel Reconciliation=true2023.06 | 9,330,000 | 2,190.6 | 91.78 | 4.15 | |
| PatchTST2023.06 | 1,174,000 | 3,842.4 | 260.367 | 8.24 | |
| CI-TSMixer (G,H)Gated Attention=true, Hierarchy Reconciliation=true2023.06 | 400,000 | 2,164.8 | 81.77 | 3.89 | |
| CI-TSMixerChannel Independence=true2023.06 | 348,000 | 1,649.4 | 65.25 | 3.03 | |
| AGCRN2023.11 | 749 | 113.46 | — | — | |
| MTGNN2023.11 | 707 | 169.34 | — | — | |
| GraphWaveNet2023.11 | 280 | 105.38 | — | — | |
| FourierGNN2023.11 | 190 | 99.25 | — | — | |
| StemGNN2023.11 | 1 | 185.86 | — | — |