Table 4. The RMSE (in kJ), mean percentage errors, and concordance correlation coefficients (CCC) of the GPR, KNNR, and LINR algorithms for the three feature types.
The figures in bold denote the smallest RMSE, percentage errors, or the largest concordance correlation coefficients.
| Algorithm | 1D | 6D | 7D | ||||||
|---|---|---|---|---|---|---|---|---|---|
| RMSE | %error | CCC | RMSE | % error | CCC | RMSE | % error | CCC | |
| GPR | 9.759 | 46.00% | 0.827 | 8.276 | 32.61% | 0.882 | 8.384 | 35.64% | 0.879 |
| KNNR | 10.163 | 42.53% | 0.813 | 8.344 | 31.84% | 0.876 | 8.415 | 29.76% | 0.847 |
| LINR | 10.596 | 41.63% | 0.833 | 9.975 | 33.65% | 0.864 | 10.229 | 29.39% | 0.847 |