Support Vector Machine-based QSPR for the Prediction of Van der Waals' Constants
Feng Luan, Ruisheng Zhang, Xiaojun Yao, Mancang Liu, Zhide Hu, Botao Fan · QSAR & Combinatorial Science · 2004
The support vector machine (SVM), as a novel type of learning machine, for the first time, was used to develop a Quantitative Structure-Property Relationship (QSPR) model of the van der Waals' constants of a diverse set of 364 compounds based on the molecular descriptors calculated from the structure alone. Multiple linear regression (MLR) was utilized to select the molecular descriptors and construct the linear model. The mean square error (MSE) errors in van der Waals' constants predictions for the whole data set given by MLR and SVM were 7.32 and 6.64 for constant a, and 3.14×10−4, 2.32×10−4 for constant b. The prediction results are in agreement with the experimental value, also, the results reveal the superiority of the SVM over MLR model.