CMStatistics 2022: Start Registration
View Submission - CMStatistics
B0804
Title: Inference for a Cauchy-type model for cylindrical data Authors:  Arthur Pewsey - University of Extremadura (Spain) [presenting]
Shogo Kato - Institute of Statistical Mathematics (Japan)
Abstract: Cylindrical data arise when the values taken by a linear variable and a circular variable are jointly observed, and consequently abound in numerous scientific disciplines. We consider inference for an appealing unimodal model for such data whose: density can be expressed in a simple closed form involving no integrals, infinite sums or special functions; parameters have clear interpretations; marginal and conditional distributions are all either Cauchy or wrapped Cauchy. For parameter estimation, we suggest a combined method of moments and robust approach and numerically-based maximum likelihood estimation, both of which are found to be computationally fast. We also propose a successful model validation tool, permutation tests for independence between the two variables, and a parametric bootstrap goodness-of-fit test. Results from Monte Carlo experiments designed to explore the finite sample characteristics of some of the inferential methods are presented, and the application of the proposed model and methods is illustrated in an analysis of data recorded during a three-day period up to and including the Great East Japan Earthquake which took place on 11th March 2011.