Title: Inference for toroidal models with circula densities generated by Fourier series
Authors: Arthur Pewsey - University of Extremadura (Spain) [presenting]
Abstract: Inference is considered for bivariate circular models derived using a marginal specification construction in combination with highly tractable circula densities generated through different patterns of non-zero Fourier coefficients. Specifically, a highly successful model identification tool and methods for parameter estimation and goodness-of-fit testing are introduced. The results from a numerical experiment comparing the modelling capabilities of such bivariate circula densities and five existing models are presented. The application of the different models is illustrated in an analysis of wind directions. An important advantage of our approach is that it facilitates the separate modelling of the circular marginals and a circula density in a structured sequential way. It also accommodates formal goodness-of-fit testing, an issue neglected in the literature related to the application of existing models for toroidal data.