Table 7 Statistic indicators of the nineteen machine learning methods on ozone prediction for testing data with three 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 | 3.274/3.117 | 2.132/2.233 | 0.118/0.236 | 0.955/0.916 | 0.906 | 0.01 |
Linr_l2 | 3.275/3.126 | 2.135/2.242 | 0.118/0.237 | 0.955/0.916 | 0.911 | 0.01 |
Lasso | 5.867/7.624 | 4.624/6.672 | 0.458/1.452 | 0.857/0.836 | 1.689 | 0.02 |
PLSR | 4.334/4.817 | 3.134/3.663 | 0.18/0.419 | 0.922/0.8 | 1.235 | 0.01 |
GRP_Expo | 2.21/5.826 | 1.506/3.593 | 0.077/0.447 | 0.98/0.713 | 6.95 | 9.7 |
GRP_DotProd | 3.274/3.121 | 2.133/2.235 | 0.118/0.235 | 0.955/0.916 | 0.909 | 15.86 |
GRP_Matern | 0.0/20.846 | 0.0/17.854 | 0.0/1.0 | 1.0/0.0 | inf | 21.98 |
SVR_linear | 3.351/2.979 | 2.053/1.957 | 0.112/0.209 | 0.953/0.924 | 0.79 | 0.01 |
SVR_poly | 3.859/5.315 | 2.636/3.747 | 0.22/0.637 | 0.938/0.804 | 1.897 | 6.99 |
SVR_rbf | 3.045/3.13 | 1.952/2.207 | 0.117/0.288 | 0.962/0.916 | 1.057 | 6.32 |
SVR_sigmoid | 6.106/6.862 | 4.568/5.585 | 0.35/1.179 | 0.845/0.6 | 1.263 | 11.4 |
MLP_1 | 3.24/3.185 | 2.105/2.275 | 0.125/0.252 | 0.956/0.913 | 0.966 | 0.37 |
MLP_2 | 15.514/21.658 | 12.678/19.454 | 1.284/4.221 | 0.0/0.0 | 1.949 | 1.39 |
RF | 3.024/3.292 | 1.977/2.323 | 0.113/0.266 | 0.962/0.907 | 1.185 | 0.05 |
Bagging | 1.346/3.453 | 0.833/2.433 | 0.045/0.264 | 0.992/0.897 | 6.581 | 0.17 |
GBoost | 3.033/3.42 | 1.987/2.58 | 0.115/0.274 | 0.962/0.902 | 1.271 | 0.3 |
AdaBoost | 4.629/5.911 | 3.663/5.04 | 0.311/0.92 | 0.918/0.855 | 1.631 | 0.17 |
HistGBoost | 2.722/3.425 | 1.813/2.53 | 0.101/0.273 | 0.969/0.9 | 1.583 | 2.35 |
LightGBM | 6.371/8.558 | 5.065/7.574 | 0.514/1.652 | 0.831/0.808 | 1.804 | 0.01 |