摘要
采用支持向量机回归(SVR)方法研究了39个麻醉药毒性的定量构效关系,基于留一法交叉验证的结果,模型的相关系数为0.970。结果表明,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(MLR)和偏最小二乘法(PLS)所得的结果。
A quantitative structure-property relationships (QSPR) study based on support vector regression (SVR) for the toxicities (pEC50) of 39 narcotics was performed. SVR method with leave-one-out cross-validation was used for evaluating the regression models. The correlation coefficient obtained by the models was 0.970. The result showed that the prediction accuracy of SVR model was higher than those of back propagation artificial neural network (BP ANN), multiple linear regression (MLR) and partial least squares (PLS) methods.
出处
《计算机与应用化学》
CAS
CSCD
北大核心
2007年第12期1617-1621,共5页
Computers and Applied Chemistry
基金
自然科学基金资助项目。(No.20503015)
关键词
定量结构性质关系
支持向量机
麻醉药毒性
Quantitative structure-property relationships, Support vector regression, narcotic toxicity
