摘要
利用股票价格波动时间序列的相关特性,基于同步理论研究股票网络的社团结构。通过对关联矩阵的谱分析确定股票网络中存在复杂的社团结构。随后,利用基于Kuramoto模型的同步聚类算法对网络节点(股票)进行动态分组,由局部序参量确定算法的收敛性并得到稳定的社团结构。通过与快速社团检测算法的对比验证,表明基于Kuramoto模型的同步聚类算法能够正确得到股票网络的社团结构,且更符合股票的属性分类。
The substructure of stock chronization theory by utilizing the network, e.g. community, is investigated based on the syn correlation matrix of time series of stock price fluctuation. Through the spectral analysis on the correlation matrix, it＇s determined that the complicated community structure obviously exists in the stock network. Then, the clustering algorithm based on the synchronization theory incorporating with the Kuramoto model is used to dynamically iden- tify the community structure, which suggests that the groups of stocks well agree with the taxon- omy of stock market. We also apply the fast community detecting algorithm to verify the former results with respect to the same parametric constrains.
出处
《复杂系统与复杂性科学》
EI
CSCD
北大核心
2014年第4期48-53,共6页
Complex Systems and Complexity Science
基金
国家自然科学基金(61004102)
中央高校基本科研业务费专项基金(ZYGX2012J075)
关键词
同步理论
股票网络
社团结构
社团检测
synchronization theory
stock network
community structure
community detection
