支持向量机方法预测有机物的亨利常数

Support vector machine applied to predict Henry's constants of organic compounds

以有机物摩尔体积V、偶极项π*、氢键给予体的酸性am、氢键接受体的碱性βm等四种理化参数为输入变量,利用支持向量机方法对72种有机物的亨利常数值进行了定量预测研究。研究发现,采用支持向量机方法可以实现使用较少样本数据建模,并达到较好的预测结果。支持向量方法的预测结果远优于线性回归法预测结果。

Support vector machine was used to predict Henry＇s constants of 72 organic compounds based on molar volume V, dipolar item π, acidity of hydrogen-bond donor am and alkalescence of hydrogen-bond acceptor βm. The results show that SVM can use fewer samples to set up model and get good output, The result obtained by SVM is better than regression of linear regression.