A comparison of the fragment constant and molecular connectivity indices models for normalized sorption coefficient estimation
Fragment constant and molecular connectivity indices models were developed based on the measured normalized sorption coefficient values of 592 chemicals. The two models were compared for precision in terms of their adjusted coefficients of determination and their mean residuals. Model robustness was...
| Veröffentlicht in: | Water environment research : a research publication of the Water Environment Federation. - 1998. - 73(2001), 3 vom: 15. Mai, Seite 307-13 |
|---|---|
| 1. Verfasser: | |
| Weitere Verfasser: | , , |
| Format: | Aufsatz |
| Sprache: | English |
| Veröffentlicht: |
2001
|
| Zugriff auf das übergeordnete Werk: | Water environment research : a research publication of the Water Environment Federation |
| Schlagworte: | Journal Article Research Support, Non-U.S. Gov't Soil Pollutants Xenobiotics |
| Zusammenfassung: | Fragment constant and molecular connectivity indices models were developed based on the measured normalized sorption coefficient values of 592 chemicals. The two models were compared for precision in terms of their adjusted coefficients of determination and their mean residuals. Model robustness was evaluated and compared using jackknifed mean residuals derived from class-by-class deletion. Results indicated that a relatively greater precision could be achieved using the fragment constant approach compared with the molecular connectivity indices model. The adjusted coefficients of determination of the two models were 0.965 and 0.765 and the mean residuals were 0.366 and 0.440, respectively. Although there were a relatively small number of independent variables involved, the robustness of the molecular connectivity indices model was greater than that of the fragment constant method as far as extrapolation among chemical classes is concerned |
|---|---|
| Beschreibung: | Date Completed 25.10.2001 Date Revised 21.09.2019 published: Print Citation Status MEDLINE |
| ISSN: | 1554-7531 |