Table 5 Performance degradation across dataset sizes for RMSE.
From: Scalable earthquake magnitude prediction using spatio-temporal data and model versioning
Algorithm | RMSE (100%) | RMSE (65%) | RMSE (35%) | RMSE (20%) | RMSE degradation (%) |
|---|---|---|---|---|---|
LightGBM | 0.535 | 0.529 | 0.544 | 0.591 | 10.47 |
XGBoost | 0.568 | 0.568 | 0.566 | 0.604 | 6.34 |
Gradient Boosting | 0.544 | 0.525 | 0.539 | 0.568 | 4.41 |
Random Forest | 0.553 | 0.563 | 0.563 | 0.603 | 9.04 |
Linear Regression | 0.691 | 0.676 | 0.707 | 0.671 | -2.89 |
