Research Article
Selection of Machine Learning Models for Oil Price Forecasting: Based on the Dual Attributes of Oil
Table 9
Error with different window sizes (future prices).
| ā | 40 | 60 | 80 | MAE | RMSE | MAE | RMSE | MAE | RMSE |
| Panel A | PCA-RNN | 0.0471 | 0.0573 | 0.0595 | 0.0703 | 0.0385 | 0.0497 | MDS-RNN | 0.0736 | 0.0871 | 0.0494 | 0.0626 | 0.1385 | 0.1529 | LLE-RNN | 0.0319 | 0.0411 | 0.0344 | 0.0442 | 0.0337 | 0.0443 | RNN | 0.0789 | 0.0940 | 0.0844 | 0.0982 | 0.1918 | 0.2217 |
| Panel B | PCA-LSTM | 0.1117 | 0.1332 | 0.1428 | 0.1649 | 0.0643 | 0.0798 | MDS-LSTM | 0.1234 | 0.1488 | 0.0798 | 0.1016 | 0.1495 | 0.1747 | LLE-LSTM | 0.0378 | 0.0491 | 0.0494 | 0.0636 | 0.0584 | 0.0749 | LSTM | 0.1336 | 0.1500 | 0.0905 | 0.1118 | 0.0715 | 0.0903 |
|
|