Research Article
Selection of Machine Learning Models for Oil Price Forecasting: Based on the Dual Attributes of Oil
Table 8
Errors of benchmark and variant model.
| ā | FWTI | SWTI | MAE | RMSE | MAE | RMSE |
| Panel A | PCA-RNN | 0.0595 | 0.0703 | 0.0600 | 0.0723 | MDS-RNN | 0.0494 | 0.0626 | 0.0649 | 0.0794 | LLE-RNN | 0.0344 | 0.0442 | 0.0323 | 0.0419 | RNN | 0.0844 | 0.0982 | 0.0845 | 0.0969 |
| Panel B | PCA-LSTM | 0.1428 | 0.1649 | 0.1269 | 0.1461 | MDS-LSTM | 0.0798 | 0.1016 | 0.0929 | 0.1126 | LLE-LSTM | 0.0494 | 0.0636 | 0.0448 | 0.0587 | LSTM | 0.0905 | 0.1118 | 0.0956 | 0.1157 |
| Panel C | PCA-BP | 0.0830 | 0.1028 | 0.0694 | 0.0864 | MDS-BP | 0.1438 | 0.1858 | 0.1008 | 0.1256 | LLE-BP | 0.0495 | 0.0621 | 0.0495 | 0.0622 | BP | 0.1247 | 0.1509 | 0.1499 | 0.1734 |
| Panel D | PCA-SVM | 0.3071 | 0.3634 | 0.2917 | 0.3578 | MDS-SVM | 0.3010 | 0.3602 | 0.2958 | 0.3686 | LLE-SVM | 0.2719 | 0.3600 | 0.2755 | 0.3679 | SVM | 0.2784 | 0.3496 | 0.2807 | 0.3529 |
|
|