Table 1.
Three-factors matrix used for melanin production using A. pullulans AKW, showing the actual and the predicted values of each point of Box-Behnken and ANN models.
| Run | Independent variable | Melanin (g/L) | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Actual ± SD | Box-Behnken | ANN | |||||||
| No | Pattern | Tyrosine (%, w/v) | Sucrose (%, w/v) | Time (h) | Predicted | Error | Predicted | Error | |
| 1* | − − 0** | 0.1 | 3 | 168 | 5.871 ± 0.490 | 5.695 | 0.176 | 5.857 | 0.014 |
| 2* | − + 0 | 0.1 | 7 | 168 | 6.954 ± 0.426 | 6.983 | − 0.029 | 6.984 | − 0.030 |
| 3* | + − 0 | 0.5 | 3 | 168 | 5.529 ± 0.152 | 5.501 | 0.029 | 5.494 | 0.035 |
| 4 | + + 0 | 0.5 | 7 | 168 | 6.783 ± 0.227 | 6.959 | − 0.176 | 6.959 | − 0.176 |
| 5 | 0 − − | 0.3 | 3 | 120 | 8.436 ± 0.325 | 8.505 | − 0.069 | 8.375 | 0.061 |
| 6* | 0 − + | 0.3 | 3 | 216 | 3. 743 ± 0.255 | 3.878 | − 0.135 | 3.749 | − 0.006 |
| 7* | 0 + − | 0.3 | 7 | 120 | 8.341 ± 0.321 | 8.206 | 0.135 | 8.221 | 0.120 |
| 8 | 0 + + | 0.3 | 7 | 216 | 6.992 ± 0.673 | 6.923 | 0.069 | 7.074 | − 0.082 |
| 9* | − 0 − | 0.1 | 5 | 120 | 7.904 ± 0.734 | 8.011 | − 0.107 | 7.967 | − 0.063 |
| 10* | + 0 − | 0.5 | 5 | 120 | 8.075 ± 0.523 | 8.035 | 0.040 | 8.167 | − 0.092 |
| 11 | − 0 + | 0.1 | 5 | 216 | 5.149 ± 0.645 | 5.189 | − 0.040 | 5.134 | 0.015 |
| 12 | + 0 + | 0.5 | 5 | 216 | 5.054 ± 0.747 | 4.947 | 0.107 | 4.955 | 0.099 |
| 13* | 000 | 0.3 | 5 | 168 | 9.804 ± 0.850 | 10.184 | − 0.380 | 10.177 | − 0.373 |
| 14* | 000 | 0.3 | 5 | 168 | 10.811 ± 0.539 | 10.184 | 0.627 | 10.177 | 0.634 |
| 15* | 000 | 0.3 | 5 | 168 | 9.973 ± 0.412 | 10.184 | − 0.247 | 10.177 | − 0.240 |
*Ten runs were used for the training process, whereas the other 5 runs were used for the validation process.
**The pattern represents the level composition of each variable within each of the tested runs, where + is the high level, − is the low level, and 0 is the middle concentration.