支持向量回归方法用于1,4,2-二氮磷杂环戊-5-(硫)酮类化合物除草活性的QSPR研究(英文)

Using support vector regression for QSPR of herbicidal l,4,2-diazaphospholidin-5-(thi) one-2-oxides

采用支持向量回归方法研究了1,4,2-二氮磷杂环戊-5-(硫)酮类化合物除草活性的QSAR。基于留一法交叉验证的结果,比较了支持向量机回归(SVR)与几种常用建模方法对于该类化合物除草活性的预测精度。研究表明:所建SVR模型的精度高于逆传播人工神经网络(BPANN)、多元线性回归和偏最小二乘(PLS)所得结果。

In the present work, QSPR of 1, 4, 2-diazaphospholidin-5-（thi） one-2-oxides with 31 compounds was analyzed by using support vector regression （SVR）. In a benchmark test, the support vector regression （SVR） models for the activity index （D） were compared to several techniques of machine learning widely used in the field. The prediction accuracies of models were discussed on the basis of the leave-one-out cross-validation （LOOCV）. The results showed that the prediction accuracy of SVR model was higher than those of back propagation artificial neural network （BPANN）, multiple linear regression （MLR） regression and partial least squares （PLS） methods.