论文部分内容阅读
针对数据波动剧烈时,一组特定的支持向量机回归参数无法满足随数据分布而改变的要求,导致回归曲线达不到所要求的精度的问题,同时针对如何有效删除在回归过程中某些非必要的数据以加快求解速度的问题,本文提出一种向量预选取的分段支持向量机回归算法.该算法首先根据数据空间分布特点删除一些非必要数据,然后根据不同区域样本的复杂程度对区间进行分段,针对各个区域设置相应的参数.仿真实验证明:p-p-SVR算法在保持回归精度的同时,较传统方法具有更好的泛化性能.
Aiming at the severe data fluctuation, a certain set of SVM regression parameters can not meet the requirements of the data distribution changes, leading to the problem that the regression curve can not reach the required accuracy. At the same time, The necessary data to speed up the solution of the problem, this paper presents a vector pre-selected piecewise support vector machine regression algorithm based on the spatial distribution of data to delete some non-essential data, and then according to the complexity of different regions of the interval The corresponding parameters are set for each region.The simulation results show that the pp-SVR algorithm has better generalization performance than traditional methods while maintaining the accuracy of regression.