Table 2.
Comparison with other methods.
| Method | MAE | RMSE | PCC |
|---|---|---|---|
| SVR | 6.38 ± 0.72 | 7.90 ± 0.70 | 0.39 ± 0.08 |
| GPR | 6.55 ± 0.72 | 8.24 ± 0.80 | 0.40 ± 0.11 |
| RFR | 6.14 ± 0.65 | 7.59 ± 0.59 | 0.43 ± 0.11 |
| LR | 6.03 ± 0.65 | 7.46 ± 0.64 | 0.47 ±0.10 |
| Alexnet | 5.98 ± 0.59 | 7.57 ± 0.72 | 0.42 ± 0.10 |
| AE | 7.65 ± 1.04 | 10.04 ± 1.07 | 0.25 ± 0.14 |
| Our proposed model using data before the Combat | 6.68 ± 0.52 | 8.48 ± 0.72 | 0.35 ± 0.13 |
| Our proposed model using data after the Combat | 5.92 ±0.62 | 7.56 ±0.78 | 0.44 ± 0.11 |
SVR, support vector regression; GPR, Gaussian process regression; RFR, random forest regression; LR, least absolute shrinkage and selection operator (LASSO) regression; AE, autoencoder; MAE, mean absolute error; RMSE, root mean square error; PCC, Pearson Correlation Coefficient.