Table 5 ML models hyperparameters and split of training and test with RMSE on 16 features dataset.
Model | Hyperparameters (16 Features) | Top-16 features | |
|---|---|---|---|
Training RMSE mm/year | Test RMSE mm/year | ||
Decision tree regression | Maximum depth = 5, Minimum samples split = 10 | 2.43 | 2.79 |
K neighbors’ regression | Nearest neighbours: 5 | 2.21 | 2.69 |
Support vector regression | C = 1, epsilon = 0.08, kernel = rbf | 1.13 | 1.77 |
Random forest regression | Number of estimators = 100, maximum depth = 5 | 1.74 | 1.98 |
FDRL | RFR: number of estimators = 100, maximum depth = 5 DTR: Maximum depth = 5, Minimum samples split = 10 KNR: Nearest neighbors = 5 SVR: C = 1, epsilon = 0.08, kernel = rbf | 1.11 | 1.32 |
