Inference on time series based on change points
Hong Wang, Jun Zhang, Hongrui Zhao · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2007
Change point problem is studied in this paper and a statistical inference method is also proposed which can be used to infer whether change points exist, how many change point there are, which kind they are and where they are. A fact is that change point theory is aimed to solve some problems of nonlinear data processing by statistics. This paper establishes a new algorithm based on Artificial Neural Network (ANN) which has self-organizing feature map (SOM) compared with the conventional approach to analyze change point and change degree. Change point can be applied to segment the phases of time series.