In this paper,based on classical linear discriminant analysis（LDA） and soft-margin support vector machine（C-SVM）,discriminant analysis via support vectors（SVDA） and margin maximizing discriminant analysis via support vectors（MSVDA） are presented.Experiments on Wine data and Iris data taken from UCI database are performed to test and evaluate the effectiveness and practicality of SVDA and MSVDA for classification problems.The results show that,SVDA and MSVDA are overall superior than LDA.%在经典线性判别分析（LDA）和软间隔支持向量机（C-SVM）的基础上,提出了支持向量判别分析（SVDA）和基于支持向量的极大化间隔判别分析（MSVDA）.为了说明SVDA和MSVDA对分类问题的有效性和实用性,利用UCI数据库中的Wine数据和Iris数据进行了对比实验.实验结果表明,总体上,SVDA和MSVDA均比LDA有效.
展开▼
