Table 4. Determining the weights in the integrated method of Hybrid+F65 by genetic algorithms.
| Generation | WHybrid | WF65 | r A | r B |
| 100 | 0.179 | 0.814 | 0.670 | 0.569 |
| 500 | 0.515 | 0.480 | 0.669 | 0.567 |
| 1000 | 0.365 | 0.631 | 0.671 | 0.570 |
| 2000 | 0.359 | 0.637 | 0.671 | 0.570 |
WHybrid belongs to the characteristic method of Hybrid, and WF65 belongs to the characteristic method of F65 in the integrated method. r A is the correlation coefficient for Hybrid+F65 trained with dataset A, and r B was validated with dataset B. Additionally, for the genetic algorithms, the population was 100 and the rates of one-point crossover and mutation were 0.7 and 0.001, respectively.