Classification of the fragrance properties of chemical compounds based on support vector machine and linear discriminant analysis

摘要: Classification models of the fragrance properties chemical compounds were performed using linear and non-linear models. The dataset was divided into three classes on basis their fragrances: apple, pineapple rose. three-class problem first explored by a classifier approach, discriminant analysis (LDA). A more accurate prediction model, machine-learning technique, support vector machine (SVM), subsequently investigated. Descriptors calculated from molecular structures alone used to represent characteristics compounds. model containing four descriptors founded SVM showed better predictive ability than LDA. accuracy in for datasets 96.6%, 80.0% 100% SVM, respectively. results indicate that can be as powerful modelling tool QSAR studies selected fragrances these Copyright (C) 2008 John Wiley & Sons, Ltd.