论文部分内容阅读
通过分子顶点类型及连接方式,构建了分子顶点特征值用于表征122个烷基苯类化合物的结构,采用多元线性回归(MLR)和偏最小二乘回归(PLS)2种方法构建了烷基苯类化合物结构与保留关系(QSRR)模型。建立的MLR模型及PLS模型的复相关系数(R)分别为0.988和0.987。两模型的估计能力和预测能力经过内部和外部双重检验,对于内部交互校验(CV)的相关系数(R_(cv))均为0.986;对于外部预测的相关系数(R_(test))分别为0.990和0.991。结果表明该构建的顶点特征值有较强的分子结构表达能力,模型具有良好的预测能力与稳定性。这对于有机化合物的定量结构保留关系研究具有一定的参考价值。
The molecular apexes were constructed to characterize the structure of 122 alkyl benzene compounds by molecular apex types and connections. Two methods, multivariate linear regression (MLR) and partial least squares regression (PLS), were used to construct the alkyl Benzene Compounds Structure-Retention Relationship (QSRR) Model. The complex correlation coefficient (R) between MLR model and PLS model was 0.988 and 0.987, respectively. The correlation coefficients (R cv) of the two models are both 0.986 for the internal cross-validation (CV) and the correlation coefficients (R test) for the external models are 0.990 and 0.991. The results show that the constructed eigenvalue has a strong ability of molecular structure expression, the model has good predictive ability and stability. This has certain reference value for the study of the quantitative structure retention of organic compounds.