论文部分内容阅读
本实验室新近提出按氢分类分子电性距离矢量(H-MEDV),用于对110个烷基苯气相色谱保留指数和54个二取代苯液相色谱容量因子进行结构表征,运用多元线性回归(multiple linear regression,MLR)以建立定量结构色谱保留关系(QSRR)模型,同时利用逐步回归结合统计检测对模型变量作筛选,建模计算值复相关系数(R_(cum))、留一法(leave-one-out,LOO)交互校验(cross-validation,CV)复相关系数(Q_(LOO)),对上述样本分别为0．9950、0．9937和0．9648、0．9530。研究结果表明,H-MEDV能较好表征该类分子结构信息,值得进一步推广。
Recently, we proposed a hydrogen-classified molecular electrical distance vector (H-MEDV) for the structural characterization of 110 alkylbenzene gas chromatographic retention indices and 54 di-substituted benzene liquid chromatographic capacity factors. Using multiple linear regression (multiple linear regression, MLR) to establish quantitative structure-retention relationship (QSRR) model, using stepwise regression combined with statistical testing to make model variables screening, modeling and calculation of complex correlation coefficient (R_ (cum) leave-one-out, LOO) cross-validation (CV) complex correlation coefficient (Q_ (LOO)) were 0.9950,0.9937 and 0.9648,0.9530 respectively. The results show that H-MEDV can better characterize the molecular structure information, which is worth further promotion.