Skip to main content

Table 6 Classification with the CAESAR model adding a 0.5 log units offset.

From: Assessment and validation of the CAESAR predictive model for bioconcentration factor (BCF) in fish

Training set

Observed logBCF

First validation set

Observed logBCF

Second validation set

Observed logBCF

327 comp.

nB

B

vB

81 comp.

nB

B

vB

119 comp.

nB

B

vB

Predicted logBCF

nB

73.70

(241)

0.31

(1)

0.00

(0)

Predicted logBCF

nB

77.78

(63)

0.00

(0)

0.00

(0)

Predicted logBCF

nB

81.51

(97)

1.68

(2)

0.00

(0)

 

B

8.87

(29)

3.06

(10)

0.31

(1)

 

B

11.11

(9)

3.70

(3)

0.00

(0)

 

B

7.56

(9)

1.68

(2)

0.84

(1)

 

vB

2.14

(7)

3.36

(11)

8.26

(27)

 

vB

1.23

(1)

2.47

(2)

3.70

(3)

 

vB

0.84

(1)

2.52

(3)

3.36

(4)

  1. Three sets are reported: training, first validation and second validation set. The percentage of the total of compounds predicted is given without considering those that are outside the applicability domain. In brackets, the number of compounds for each class. The total number of compounds is also reported. To take account of the endpoint variability, the predicted values are modified adding an offset of 0.5 log units for the compounds near the B and vB thresholds.