SVM用于2-杂环芳基苯并二氢吡喃-4-酮衍生物QSRR研究

Support Vector Machine Applied to Study on Quantitative Structure-Rentention Relationships of 2-Heteroaryl-4-Chromanone Derivatives

将支持向量机(SVM)用于15种2-杂环芳基苯并二氢吡喃-4-酮衍生物的定量结构-色谱保留相关(QSRR)研究.通过核函数的选择及核函数参数的优化,建立了预测模型,预测了该类化合物色谱容量因子,得到优于多元线性回归(MLR)方法的预测结果.实践表明,SVR算法能较好地解决小样本、非线性等问题,并能够有效控制过拟合,提高算法的预报能力.

Support vector machine is suitable for the data processing based on finite number of training samples, with special technique to restrict overfitting. In this work, support vector regression （SVR） algorithm was applied to quantitative structure - rentention relationships study on 2-heteroaryl-4-chromanone derivatives. A more predictive model was set up to predict the retention property of the title compounds through optimizing kernel function and its parameters. The modeling result obtained by SVR is better than that of multiple linear regression （ MLR ）.