论文部分内容阅读
应用按氢分类的分子电性距离矢量(H-MEDV)对圆角桉叶精油的51种组分的气相色谱保留时间(t_R)进行了定量结构-色谱保留关系研究。通过多元线性回归,得到模型(M1)的复相关系数(R_1)为0.971,标准偏差(SD_1)为2.921。用逐步回归的方法挑选了8个变量进行建模,得到的模型(M2)复相关系数(R_2)为0.970,标准偏差(SD_2)为2.880。再用留一法(Leave-one-out,LOO)交互检验对两模型进行评价,得到两模型的复相关系数(R_(CV))分别为0.947和0.952,结果表明所建模型具有较好的稳定性和预测能力。
The gas chromatographic retention time (t_R) of 51 components of Eucalyptus urophylla oil was quantitatively studied by using H-MEDV (hydrogen-classified molecular electrical distance vector). By multivariate linear regression, the complex correlation coefficient (R_1) of the model (M1) was 0.971, and the standard deviation (SD_1) was 2.921. Eight variables were selected by stepwise regression method to model the complex correlation coefficient (R2) of the model (M2) 0.970, and the standard deviation (SD_2) was 2.880. Then the leave-one-out (LOO) interaction test was used to evaluate the two models, and the complex correlation coefficient (R CV) of the two models was 0.947 and 0.952 respectively. The results showed that the proposed model had good Stability and predictive ability.