Modeling the Complex Network of Multidimensional Information Time Series to Characterize the Volatility Pattern Evolution

Time series processing and analyzing is one of the major challenges in big data research, especially in inferring the dynamical mechanism of time series. In this paper, we constructed a complex network from time series via exploring the evolutionary relationship among the volatility patterns. We introduced the symbolic method and sliding window to describe the volatility patterns of time series that contains multidimensional information. Meanwhile, we explored the evolutionary mechanism of these volatility patterns based on the topological characteristics in network. In our research, we selected six stock indices around the world as sample data. Interestingly, for the six networks, they all showed a “petal-shaped”structure which consists of a core and loops. Moreover, through analyzing the topological characteristics of the six networks, we discovered distinguished results of their overall characteristics and loop length distributions. Furthermore, we uncovered the media patterns which trigger the structure of network changing among core and loops. In a word, this paper described the volatility patterns and explored the topological characteristics in networks constructed from time series data, which provides a novel perspective to understand the evolutionary dynamic mechanism.