Predicting of BCFs in Fish for Organic Chemicals Using Fragment Constant Metho D

胡海瑛,陶澍,卢晓霞
DOI: https://doi.org/10.3321/j.issn:0253-2468.2001.03.004
2001-01-01
Abstract:Based on 337 experimentally determined BCF values in fish for nonpolar and polar chemicals,a fragment constant estimation model for log BCF was established.In the first step of regression process,the whole body of the database was divided into two parts,one containg 277 chemicals was used for model development and the other of 60 chemicals for model validation.As a result,an average error of 0.320 log unit was produced when prediction of log BCF was made to the test set chemicals by applying the regression model derived from the train set data.The two sets of data then were combined to generate the final prediction model,whose determination coefficient was 0.982,with a standard error of 0.453 and a mean absolute residual of 0.322 log unit.
What problem does this paper attempt to address?