论文部分内容阅读
由反应物和产物的结构衍生了反应物、产物和化学反应的MOLMAP指数,其中化合物的结构由化学键的物理化学性质和拓扑性质所表征.将前述MOLMAP指数应用于一个含七类光化学反应的数据集,通过随机森林建立了三种类型的模型:(1)预测反应物可能发生的反应类型;(2)预测可能合成产物的反应类型;(3)预测整个化学反应的类型.由于无需指定数据集中参与反应的化学键,所以,MOLMAP指数能够得到广泛的应用.所得分类预测结果好于我们此前对同一数据集的研究,表明改进化学键的描述有助于提高MOLMAP指数的预测能力.
The MOLMAP indices of reactants, products and chemical reactions were derived from the structure of the reactants and products, and the structure of the compounds was characterized by the physicochemical and topological properties of the chemical bonds.The MOLMAP index mentioned above was applied to a data containing seven photochemical reactions Set of three types of models through random forests: (1) predict the types of reactions that may occur with the reactants; (2) predict the types of reactions that are likely to synthesize; (3) predict the type of chemical reaction as a whole Therefore, the MOLMAP index can be widely used.The results of the classification are better than our previous research on the same data set, indicating that improving the description of the chemical bond can improve the prediction ability of the MOLMAP index.