Prediction of the melting point temperature using a linear QSPR for homologous series

Inga Paster, Mordechai Shacham, Neima Brauner

Research output: Chapter in Book/Report/Conference proceedingChapterpeer-review

1 Scopus citations


Prediction of normal melting temperature (Tm) using linear Quantitative Structure Property Relationships (QSPR) whose applicability domain is limited to a particular homologous series is considered. It is shown that by limiting the applicability domain of the QSPR and using a very large bank of descriptors it is possible to identify a small set of descriptors whose linear combination represents Mm within experimental error level. even if the change of Tm, with the number of C atoms is highly irregular. Confidence in the predicted values in both interpolation and extrapolation is considerably enhanced by ensuring random residual distribution in the training set used. The proposed method yielded prediction errors lower than reported in the literature in all the homologous series that were included in this study.

Original languageEnglish
Title of host publication18th European Symposium on Computer Aided Process Engineering
EditorsBertrand Braunschweig, Xavier Joulia
Number of pages6
StatePublished - 3 Oct 2008

Publication series

NameComputer Aided Chemical Engineering
ISSN (Print)1570-7946


  • QSPR
  • homologous series
  • melting point
  • molecular descriptors
  • property prediction

ASJC Scopus subject areas

  • General Chemical Engineering
  • Computer Science Applications


Dive into the research topics of 'Prediction of the melting point temperature using a linear QSPR for homologous series'. Together they form a unique fingerprint.

Cite this