Table 6 Statistic indicators of the nineteen machine learning methods on ozone prediction for testing data with one-time-lagged ozone values
From: A comparison of machine learning methods for ozone pollution prediction
Model | RMSE \(\downarrow\) | MAE \(\downarrow\) | MAPE \(\downarrow\) | R-square \(\uparrow\) | J2 \(\downarrow\) | Time (min) |
|---|---|---|---|---|---|---|
Linr | 3.315/3.217 | 2.171/2.289 | 0.12/0.248 | 0.954/0.911 | 0.942 | 0.01 |
Linr_l2 | 3.315/3.22 | 2.173/2.293 | 0.121/0.249 | 0.954/0.911 | 0.944 | 0 |
Lasso | 5.867/7.614 | 4.624/6.66 | 0.458/1.448 | 0.857/0.836 | 1.684 | 0.01 |
PLSR | 4.987/6.558 | 3.849/5.405 | 0.213/0.63 | 0.897/0.629 | 1.729 | 0.01 |
GRP_Expo | 2.685/4.166 | 1.797/3.086 | 0.097/0.414 | 0.97/0.86 | 2.407 | 7.79 |
GRP_DotProd | 3.315/3.215 | 2.17/2.286 | 0.12/0.249 | 0.954/0.911 | 0.941 | 15.59 |
GRP_Matern | 0.0/20.863 | 0.0/17.878 | 0.0/1.0 | 1.0/0.0 | inf | 21.61 |
SVR_linear | 3.388/3.032 | 2.087/1.979 | 0.114/0.213 | 0.952/0.921 | 0.801 | 0.01 |
SVR_poly | 3.903/5.351 | 2.678/3.778 | 0.222/0.624 | 0.937/0.801 | 1.88 | 5.37 |
SVR_rbf | 3.097/3.189 | 1.998/2.277 | 0.12/0.298 | 0.96/0.914 | 1.06 | 6.04 |
SVR_sigmoid | 4.501/4.079 | 3.352/3.241 | 0.215/0.521 | 0.916/0.86 | 0.821 | 6.92 |
MLP_1 | 3.315/3.199 | 2.174/2.234 | 0.126/0.232 | 0.954/0.912 | 0.931 | 0.28 |
MLP_2 | 15.514/21.707 | 12.668/19.503 | 1.287/4.222 | 0.0/0.0 | 1.958 | 1.09 |
RF | 3.07/3.416 | 2.002/2.451 | 0.114/0.272 | 0.961/0.9 | 1.238 | 0.05 |
Bagging | 1.381/3.731 | 0.863/2.756 | 0.046/0.281 | 0.992/0.881 | 7.299 | 0.16 |
GBoost | 3.068/3.663 | 2.001/2.719 | 0.115/0.278 | 0.961/0.885 | 1.425 | 0.29 |
AdaBoost | 4.908/6.396 | 3.942/5.485 | 0.326/0.985 | 0.912/0.844 | 1.698 | 0.12 |
HistGBoost | 2.806/3.604 | 1.855/2.667 | 0.102/0.284 | 0.967/0.889 | 1.65 | 0.89 |
LightGBM | 6.568/8.834 | 5.241/7.857 | 0.527/1.682 | 0.821/0.804 | 1.809 | 0.01 |