Characterize System Dynamic of Pseudo Periodic Time Series with Evolution Networks

The appearing sequence of data points in time series is denoted as data evolution. Modeling data evolution is important to discover complex system dynamic. Inspired by the complex networks, in this paper, we try to map data evolution to a process of network evolution. To transform pseudo periodic time series to complex network, each cycle is regarded as a node of the network. Two different definitions of edges are discussed. One is the neighborhood of the cycles in time series, and the other is the similarity. Based on the evolution network models, data evolution model is defined and composed three steps: start, growth, preference. With a given transforming rule, the network evolution is only constrained by system dynamic. So data evolution model can help us discover hidden dynamic and understand complex system. An appropriate example about heart rate variation is analyzed to illustrate the effect of the model.