Phenol Compounds QSAR Modeling Based on Support Vector Regression

Objective: Quantitative structure-activity relationship(QSAR) study to explore the relationship of toxicity of phenol compounds and molecular structure parameters.Methods: The QSAR study phenol compounds and their derivatives based on support vector regression(SVR) and according to the mean square error minimum principle to choose the best kernel function.Results: The results suggested that different data sets need different optimal kernel functions.SVR has the better advantages of stability and strong generalization ability in the small samples and non-linear problems.In the QSAR study on phenol compounds and their derivatives,the SVR model superior to the results of literature can be gained.Conclusion: SVR model has good predictive ability.Therefore,SVR can be widely used in QSAR and other related fields.