Title: Recent cylindrical models and their applications
Authors: Toshihiro Abe - Hosei University (Japan) [presenting]
Abstract: Correlation/covariance is a fundamental concept, and multivariate analyses often begin by investigating their presence in given multivariate data. In particular, if an objective variable is not a scalar but a vector with correlations, to construct a statistical model, we need multivariate probability distributions that can flexibly capture the correlations between the objective variables. For example, the multivariate Gaussian distribution is used in the Gaussian process and the geostatistical process to express stochastic uncertainty in temporally or spatially correlated objective variables. By focusing on cylindrical distributions, we show examples of cylindrical data. After a brief review of probability distributions on the line and on the circle, we introduce WeiSSVM distribution. Using a statistical model of forest tree data in Finland, we demonstrate an application of the cylindrical distributions to quantify the factors that affect asymmetric crown expansion.