QSAR Study of Arylpyridone Oxime Based on the SVM and Elman Algorithms

Based on topological chemistry theory, a quantum chemistry method is used to calculate the 91 electrical distance vector (Mi) of a molecule, which is used to characterize the chemical microenvironment in which the molecule is located. The multivariate stepwise regression method was used to screen the variables to obtain the ternary best equation. Correlation coefficient R2=0.887, R2CV=0.673, passed by FIT and AIC .The ternary variable is used as the input set, and the inhibition rate is used as the output set. The LS-SVM and Elman-ANN algorithms are used for prediction and comparison. The results show that R2 is 0.993 and 0.994 respectively. The prediction ability is similar. But the stability of Elman is better, and the two-dimensional structure affecting the molecule is structural fragments such as =CH-, -CH<.