Table 2. Model predictions versus actual values for ten samples from the independent test set (n= 23).
| Total length (cm) | Actual maturity (0/1) | Actual age (years) a | Predicted maturity (continuous output) | Predicted maturity (0/1) b | Predicted age (years) |
|---|---|---|---|---|---|
| 16.7 | 0 | 1 | 0.0001 | 0 | 0.9874 |
| 24.2 | 0 | 2 | −0.0057 | 0 | 2.0075 |
| 26.6 | 0 | 2 | −0.0056 | 0 | 2.0079 |
| 30.4 | 1 | 3 | 0.9849 | 1 | 2.9706 |
| 37.4 | 1 | 4 | 1.0034 | 1 | 3.564 |
| 42.5 | 1 | 5 | 1 | 1 | 4.9761 |
| 43.1 | 1 | 5 | 1 | 1 | 5.0408 |
| 43.5 | 1 | 5 | 1 | 1 | 5.081 |
| 55.3 | 1 | 7 | 1 | 1 | 7.0723 |
| 62.7 | 1 | 9 | 1 | 1 | 8.9853 |
Notes.
Actual age was determined by scale reading following standard sclerochronological protocols.
Binary classification obtained by thresholding continuous output at ≥ 0.5.