论文部分内容阅读
基于 592种化合物的Koc实测数据 ,分别采用片段常数法和分子连接性指数法建立了其估算模型 .根据模型残差分析了两种模型的预测误差 ,并运用 2种方式的jackknife检验比较了它们的稳健性 .片段常数模型涉及较多变量 ,其应用受到有限的已知片段常数以及烦琐的片段划分过程的限制 ,相比之下 ,分子连接性指数模型变量数较少 ,其应用时只需了解化合物的结构式 .片段常数模型的估算精度略高于分子连接性指数模型 ,两者的差别主要体现在logKoc<1和logKoc >4的化合物上 .分子连接性指数模型的稳健性比片段常数模型高 .在依次删除一类特定类别化合物时 ,其估算误差的变异幅度均小于片段常数模型 .
Based on the Koc measured data of 592 compounds, the estimation models were established by the piecewise constant method and the molecular connectivity index method, respectively, and the prediction errors of the two models were analyzed based on the model residuals. Two methods of jackknife test were used to compare them The fragment constant model involves many variables, its application is limited by the limited known fragment constants and cumbersome fragmenting process, in contrast, the molecular connectivity index model has fewer variables and its application needs only The structural constants of the compounds are known.The estimation accuracy of the fragment constant model is slightly higher than that of the molecular connectivity index model, the difference between the two is mainly reflected in the compounds with logKoc <1 and logKoc> 4. The molecular connectivity index model is more robust than the fragment constant model The variation of the estimation errors of all the compounds in a particular category were less than the fragment constant model.