Title: Complex-valued time-series models and their relation to directional statistics
Authors: Takayuki Shiohama - Nanzan University (Japan) [presenting]
Abstract: Stationary time series fluctuation often exhibits periodic behavior and these patterns are summarized via a spectral density, which can be modeled using a circular distribution function. Several time-series models are studied in relation to a circular distribution. First, we illustrate how to model bivariate time-series data using complex-valued time series in the context of circular distribution functions. These models are then extended to have a skewed spectrum by incorporating a sine-skewing transformation. Further, two parameter estimation methods are introduced, and their asymptotic properties are investigated. These theoretical results are verified via a Monte Carlo simulation. In addition, real data analyses are performed to illustrate the applicability of the proposed models.