摘要
为提高市场偏好预警分析算法的有效性,提出一种基于灰色混合核AR-SVM模型的新方法。使用支持向量机(SVM)算法来构建财务市场风险预警分析模型,该模型存在非极端风险和极端风险两种情况。采用SVM算法找到基于训练集的最优分类过程。SVM模型容易出现极端风险预警,因此由改进的灰色模型处理市场偏好预测问题的错误市场偏好数据。采用混合核函数对SVM算法进行改进,实现样本数据,提高自回归模型的预测性能。SVM算法可以用于提高市场预警分析的准确性。实验结果表明,该方法可以很好地分析市场偏好数据。
In order to improve the effectiveness of market preference early warning analysis, we proposed a new method based on grey-scale hybrid kernel AR-SVM model. We used SVM to construct financial market risk early warning analysis model which included no extreme risk and extreme risk in two cases. We used SVM to find the optimal classification process based on training set. SVM model was prone to extreme risk early warning, so the improved grey-scale model was used to deal with the wrong market preference data of market preference prediction problem. We used hybrid kernel function to improve the SVM algorithm, realized sample data and improved the prediction performance of the auto regressive model. SVM algorithm was used to improve the accuracy of market early warning analysis. The experimental results show that the proposed method can well analyze market preference data.
作者
贾茹阁
张忠林
Jia Ruge;Zhang Zhonglin(College of Electronic and Information Engineering, Lanzhou Jiaotong University, Lanzhou 730070, Gansu, China)
出处
《计算机应用与软件》
北大核心
2019年第2期200-204,251,共6页
Computer Applications and Software
基金
国家自然科学基金项目(61662043)
