Table 3 Perform boosting by applying the proposed framework to Transformer and its variants. We report the average MSE/MAE (illustrated in Table 2) and the relative MSE reduction rate (Promotion) for all the predicted lengths in our framework.
From: NSPLformer: exploration of non-stationary progressively learning model for time series prediction
Dataset | Solar Energy | Electricity | Traffic | Weather | ||||
|---|---|---|---|---|---|---|---|---|
MSE | MAE | MSE | MAE | MSE | MAE | MSE | MAE | |
Transformer | 1.427 | 0.914 | 0.275 | 0.371 | 0.714 | 0.4 | 0.659 | 0.571 |
+Ours | 0.239 | 0.264 | 0.173 | 0.264 | 0.413 | 0.272 | 0.259 | 0.281 |
Promotion | 77.18% | 32.97% | 37.08% | 55.74% | ||||
Informer | 1.546 | 0.994 | 0.312 | 0.401 | 0.763 | 0.417 | 0.633 | 0.552 |
+Ours | 0.197 | 0.228 | 0.167 | 0.261 | 0.162 | 0.244 | 0.245 | 0.271 |
Promotion | 82.16% | 40.69% | 60.13% | 56.10% | ||||
Autoformer | 0.427 | 0.399 | 0.225 | 0.336 | 0.627 | 0.378 | 0.339 | 0.379 |
+Ours | 0.391 | 0.369 | 0.192 | 0.275 | 0.427 | 0.275 | 0.266 | 0.285 |
Promotion | 7.97% | 16.41% | 29.57% | 23.17% | ||||
