From: Modeling nanofluid viscosity: comparing models and optimizing feature selection—a novel approach
Model type | Preset | RMSE (validation) | MSE (validation) | R2 (validation) | MAE (validation) | MAE (test) | MSE (test) | RMSE (Test) | R2 (test) |
---|---|---|---|---|---|---|---|---|---|
Gaussian process regression | Custom Gaussian process regression | 0.022 | 0.0005 | 0.9941 | 0.0146 | 0.0157 | 0.0004 | 0.0211 | 0.9924 |
Gaussian process regression | Custom Gaussian process regression | 0.0228 | 0.0005 | 0.9936 | 0.0129 | 0.0186 | 0.0006 | 0.0251 | 0.9894 |
Gaussian process regression | Custom Gaussian process regression | 0.0232 | 0.0005 | 0.9934 | 0.0131 | 0.0186 | 0.0006 | 0.0251 | 0.9894 |
Gaussian process regression | Custom Gaussian process regression | 0.0263 | 0.0007 | 0.9915 | 0.0144 | 0.0237 | 0.0012 | 0.034 | 0.9805 |
Gaussian process regression | Exponential GPR | 0.0302 | 0.0009 | 0.9889 | 0.0172 | 0.0201 | 0.0008 | 0.0278 | 0.987 |
Gaussian process regression | Rational quadratic GPR | 0.0341 | 0.0012 | 0.9858 | 0.0192 | 0.0203 | 0.0008 | 0.0283 | 0.9864 |
Gaussian process regression | Matern 5/2 GPR | 0.0359 | 0.0013 | 0.9842 | 0.0208 | 0.0229 | 0.0009 | 0.0298 | 0.985 |
Neural network | Medium neural network | 0.0363 | 0.0013 | 0.9839 | 0.0222 | 0.0372 | 0.0034 | 0.0585 | 0.9422 |
Neural network | Bilayered neural network | 0.0367 | 0.0013 | 0.9835 | 0.0218 | 0.0176 | 0.0005 | 0.0226 | 0.9914 |
Neural network | Trilayered neural network | 0.0388 | 0.0015 | 0.9816 | 0.023 | 0.0449 | 0.0051 | 0.0713 | 0.914 |
Gaussian process regression | Squared exponential GPR | 0.0431 | 0.0019 | 0.9773 | 0.0256 | 0.0277 | 0.0015 | 0.0393 | 0.9738 |
Neural network | Wide neural network | 0.0438 | 0.0019 | 0.9765 | 0.025 | 0.0374 | 0.0027 | 0.0521 | 0.9541 |
Neural network | Narrow neural network | 0.052 | 0.0027 | 0.9669 | 0.0293 | 0.0408 | 0.0039 | 0.0621 | 0.9348 |
Linear regression | Interactions Linear | 0.0535 | 0.0029 | 0.965 | 0.0371 | 0.0396 | 0.0025 | 0.0495 | 0.9585 |
Stepwise linear regression | Stepwise linear | 0.0554 | 0.0031 | 0.9624 | 0.0394 | 0.0388 | 0.0028 | 0.0528 | 0.9529 |
SVM | Medium Gaussian SVM | 0.0582 | 0.0034 | 0.9585 | 0.0438 | 0.0451 | 0.0026 | 0.0514 | 0.9552 |
Tree | Fine tree | 0.0592 | 0.0035 | 0.9571 | 0.0368 | 0.0328 | 0.0022 | 0.0464 | 0.9636 |
SVM | Quadratic SVM | 0.0617 | 0.0038 | 0.9534 | 0.0476 | 0.0349 | 0.0021 | 0.046 | 0.9642 |
Linear regression | Linear | 0.0642 | 0.0041 | 0.9496 | 0.0504 | 0.0416 | 0.0028 | 0.0529 | 0.9526 |
Linear regression | Robust linear | 0.0653 | 0.0043 | 0.9478 | 0.0511 | 0.0406 | 0.0026 | 0.0513 | 0.9554 |
SVM | Linear SVM | 0.0658 | 0.0043 | 0.947 | 0.0517 | 0.0411 | 0.0026 | 0.0511 | 0.9558 |
SVM | Fine Gaussian SVM | 0.0717 | 0.0051 | 0.9371 | 0.0518 | 0.0494 | 0.0037 | 0.0611 | 0.9368 |
SVM | Cubic SVM | 0.075 | 0.0056 | 0.9311 | 0.0545 | 0.0393 | 0.0038 | 0.0617 | 0.9356 |
Gaussian process regression | Custom Gaussian process regression | 0.0819 | 0.0067 | 0.9179 | 0.0389 | 0.0549 | 0.0121 | 0.1099 | 0.7955 |
Ensemble | Boosted trees | 0.0819 | 0.0067 | 0.9178 | 0.0495 | 0.0315 | 0.0019 | 0.0435 | 0.968 |
Gaussian process regression | Custom Gaussian process regression | 0.0877 | 0.0077 | 0.9058 | 0.0663 | 0.0801 | 0.0093 | 0.0965 | 0.8425 |
SVM | Coarse Gaussian SVM | 0.0945 | 0.0089 | 0.8906 | 0.0702 | 0.0604 | 0.0046 | 0.0677 | 0.9225 |
Ensemble | Bagged trees | 0.1021 | 0.0104 | 0.8724 | 0.0641 | 0.0593 | 0.0091 | 0.0954 | 0.8459 |
Kernel | SVM kernel | 0.1349 | 0.0182 | 0.7772 | 0.0947 | 0.1012 | 0.0235 | 0.1534 | 0.602 |
Kernel | Least squares regression kernel | 0.1473 | 0.0217 | 0.7343 | 0.117 | 0.1303 | 0.0247 | 0.1572 | 0.5817 |
Tree | Medium tree | 0.1504 | 0.0226 | 0.7232 | 0.0941 | 0.0546 | 0.0165 | 0.1286 | 0.7204 |
Tree | Coarse tree | 0.2349 | 0.0552 | 0.3243 | 0.1742 | 0.0967 | 0.0137 | 0.1171 | 0.7679 |