一类变量可分离的支持向量分类机的研究与应用

Research on Application of a Kind of Support Vector Classification Machine with Variables Seperable

针对传统 SVC 方法在样本容量大时存在训练时间过长的不足,建立了一种变量可分离的支持向量分类模型DCSVC,并将其应用于随机生成的模拟数据的学习与航空公司旅客运输数据的预测中。实践证明,DCSVC 算法预测的误差小于传统 SVM 算法,具有较高的精度,且训练时间比传统 SVC 短。

This paper sets up a new kind of support vector regression machine model ＂DCSVM＂ which variables can be separated. And the experiment demonstrates that the error of DCSVM is less than traditional support vector machine. We also get a good regression result in the airline passenger volumn forecast using DCSVM method.