Table 1 Assessment metrics for machine learning models: performance during all segments.
From: Data-driven frameworks to robustly predict solubility parameter of diverse polymers
Model | R² | RMSE | MRD% | ||||||
|---|---|---|---|---|---|---|---|---|---|
Training | Validation | Testing | Training | Validation | Testing | Training | Validation | Testing | |
Linear Regression | 0.751 | 0.745 | 0.763 | 2223 | 2040 | 2242 | 11.5 | 11.3 | 6.3 |
Ridge Regression | 0.751 | 0.748 | 0.763 | 2223 | 2029 | 2238 | 13.3 | 12.8 | 6.4 |
Lasso Regression | 0.751 | 0.751 | 0.765 | 2224 | 2016 | 2232 | 11.4 | 16.7 | 6.4 |
Elastic Net | 0.751 | 0.749 | 0.764 | 2223 | 2023 | 2236 | 15.6 | 14.3 | 6.3 |
Gradient Boosting | 0.987 | 0.884 | 0.888 | 503 | 1376 | 1540 | 2.0 | 4.3 | 5.9 |
Random Forest | 0.987 | 0.884 | 0.888 | 503 | 1376 | 1540 | 2.0 | 4.3 | 5.0 |
XGBoost | 1.000 | 0.827 | 0.828 | 0 | 1678 | 1910 | 0.0 | 5.1 | 4.9 |
LightGBM | 0.960 | 0.847 | 0.849 | 887 | 1577 | 1786 | 1.7 | 4.6 | 5.3 |
CatBoost | 0.961 | 0.870 | 0.893 | 875 | 1455 | 1503 | 3.3 | 4.3 | 6.2 |
SVR | 0.904 | 0.884 | 0.863 | 1382 | 1373 | 1705 | 2.7 | 3.3 | 4.4 |
KNN | 0.877 | 0.801 | 0.792 | 1560 | 1803 | 2097 | 4.2 | 5.6 | 4.0 |
Decision Tree | 1.000 | 0.744 | 0.695 | 10 | 2045 | 2540 | 0.0 | 7.2 | 3.7 |
ANN | 0.968 | 0.838 | 0.914 | 798 | 1626 | 1351 | 3.0 | 4.3 | 4.3 |
CNN | 0.941 | 0.877 | 0.917 | 1079 | 1415 | 1329 | 3.8 | 4.3 | 4.6 |
