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€‰FWTISWTI
MAERMSEMAERMSE

Panel A
PCA-RNN0.05950.07030.06000.0723
MDS-RNN0.04940.06260.06490.0794
LLE-RNN0.03440.04420.03230.0419
RNN0.08440.09820.08450.0969

Panel B
PCA-LSTM0.14280.16490.12690.1461
MDS-LSTM0.07980.10160.09290.1126
LLE-LSTM0.04940.06360.04480.0587
LSTM0.09050.11180.09560.1157

Panel C
PCA-BP0.08300.10280.06940.0864
MDS-BP0.14380.18580.10080.1256
LLE-BP0.04950.06210.04950.0622
BP0.12470.15090.14990.1734

Panel D
PCA-SVM0.30710.36340.29170.3578
MDS-SVM0.30100.36020.29580.3686
LLE-SVM0.27190.36000.27550.3679
SVM0.27840.34960.28070.3529