Table 2.
The average performance of different models in predicting absorption wavelengths in 3 independent runs
| Training set | Test set | |||||||
|---|---|---|---|---|---|---|---|---|
| Model | MAE | RMSE | MRE | R 2 | MAE | RMSE | MRE | R 2 |
| GCN | 18.6570 | 27.1291 | 0.0420 | 0.9260 | 21.1936 | 35.4761 | 0.0477 | 0.8709 |
| GIN | 24.5175 | 33.6847 | 0.0557 | 0.8796 | 23.3730 | 37.0351 | 0.0523 | 0.8535 |
| GAT | 29.8693 | 43.9367 | 0.0675 | 0.7937 | 30.9669 | 46.0222 | 0.0691 | 0.7765 |
| NGNN | 10.7600 | 15.8400 | 0.0236 | 0.9800 | 10.6500 | 18.3300 | 0.0237 | 0.9600 |
| GBRT | 6.8296 | 9.8413 | 0.0160 | 0.9905 | 14.1696 | 25.0745 | 0.0324 | 0.9368 |
| RF | 4.8454 | 9.4383 | 0.0109 | 0.9912 | 13.0689 | 27.1370 | 0.0299 | 0.9258 |