By using neural networks, correlations were established between chemical structure and boiling points of chlorofluorocarbons with 1, 1–2, or 1–4 carbon atoms (15, 62, and 276 compounds, respectively) as well as of halomethanes with up to four different halogens (48 compounds). The molecular descriptors included the number of carbon atoms and of each type of halogen atom as well as topological indices. Results were validated by the jackknifing procedure. The correlation coefficients were r = 0.985–0.995. Predictions were made for the boiling points of several haloethanes.
|Original language||English (US)|
|Number of pages||4|
|Journal||Journal of chemical information and computer sciences|
|State||Published - Sep 1 1994|