Table 7.
Classification with the BCFBAF v3.00 model.
| Training set | Observed logBCF | Validation set | Observed logBCF | External set | Observed logBCF | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 450 comp. | nB | B | vB | 103 comp. | nB | B | vB | 82 comp. | nB | B | vB | |||
| Predicted logBCF | nB | 82.00 (369) |
3.33 (15) |
2.00 (9) |
Predicted logBCF | nB | 81.55 (84) |
2.91 (3) |
1.94 (2) |
Predicted logBCF | nB | 39.02 (32) |
12.20 (10) |
3.66 (3) |
| B | 2.00 (9) |
0.67 (3) |
2.00 (9) |
B | 0.97 (1) |
0.97 (1) |
0.00 (0) |
B | 10.98 (9) |
4.88 (4) |
4.88 (4) |
|||
| vB | 2.22 (10) |
0.89 (4) |
4.89 (22) |
vB | 3.88 (4) |
0.97 (1) |
6.80 (7) |
vB | 6.10 (5) |
3.66 (3) |
14.63 (12) |
|||
Three sets are reported for the compounds of the dataset: training, validation and external. The percentage of the total of compounds predicted is given without considering those outside the applicability domain. In brackets, the number of compounds for each class is reported. The total number of compounds is also reported.