Quantitative Structure–Property Relationship for Liquid Heat Capacity of Alcohols
Volume Title: 1
گروه مهندسی شیمی، دانشگاه صنعتی بیرجند، بیرجند، ایران
In this work a new simple QSPR model with only three parameters was developed for prediction the liquid heat capacity of alcohols from their molecular structures. By using the modified particle swarm optimization (MPSO)-multiple linear regression (MLR) a subset of relevant descriptors from a large pool of descriptors was selected and a linear model is developed. The QSPR model was validated using internal and external validation methods. The MPSO-MLR model can give satisfactory prediction results with respect to squared correlation coefficient (R2) of 0.925 and root mean squares error (RMSE) of 16.847 for data set. Interpretation of the model descriptors indicated that the branching and atom composition diversity of a molecule is the most important factor affecting the liquid heat capacity of an alcohol.
Heat capacity of liquid; Alcohol; Quantitative structure property relationship (QSPR); Modified particle swarm optimization (MPSO); Multiple linear regression (MLR)