使用逐步回归分析-人工神经网络（SPA-ANN）和逐步回归分析-多元线性回归（SPA-MLR）对农药土壤吸附系数（K(OC)）进行定量结构-性质关系（QSPR）建模。

QSPR modeling of soil sorption coefficients (K(OC)) of pesticides using SPA-ANN and SPA-MLR.

作者信息

Goudarzi Nasser, Goodarzi Mohammad, Araujo Mario Cesar Ugulino, Galvão Roberto Kawakami Harrop

机构信息

Faculty of Chemistry, Shahrood University of Technology, P.O. Box 316, Shahrood, Iran.

出版信息

J Agric Food Chem. 2009 Aug 12;57(15):7153-8. doi: 10.1021/jf9008839.

DOI:10.1021/jf9008839

PMID:19722589

Abstract

A quantitative structure-property relationship (QSPR) study was conducted to predict the adsorption coefficients of some pesticides. The successive projection algorithm feature selection (SPA) strategy was used as descriptor selection and model development method. Modeling of the relationship between selected molecular descriptors and adsorption coefficient data was achieved by linear (multiple linear regression; MLR) and nonlinear (artificial neural network; ANN) methods. The QSPR models were validated by cross-validation as well as application of the models to predict the K(OC) of external set compounds, which did not contribute to model development steps. Both linear and nonlinear methods provided accurate predictions, although more accurate results were obtained by the ANN model. The root-mean-square errors of test set obtained by MLR and ANN models were 0.3705 and 0.2888, respectively.

摘要

进行了定量结构-性质关系（QSPR）研究，以预测某些农药的吸附系数。采用连续投影算法特征选择（SPA）策略作为描述符选择和模型开发方法。通过线性（多元线性回归；MLR）和非线性（人工神经网络；ANN）方法对所选分子描述符与吸附系数数据之间的关系进行建模。通过交叉验证以及将模型应用于预测外部集化合物的K(OC)来验证QSPR模型，外部集化合物未参与模型开发步骤。线性和非线性方法均提供了准确的预测，尽管ANN模型获得了更准确的结果。MLR和ANN模型获得的测试集均方根误差分别为0.3705和0.2888。