Regression Models for Directional Variables

Document Type

Book Chapter

Publication Title

Forum for Interdisciplinary Mathematics

Abstract

The statistical analysis of directional random variables requires different techniques than the linear random variables due to the difference in topology of their sample spaces. The directional random variables have many applications in diverse fields of studies such as meteorology, ecology and medical sciences. Spherical-spherical regression models provide a framework for modelling a directional response on other directional covariates. This paper provides a summary of the various models considered in this context, namely trigonometric polynomial-based model, rotational models, decentred predictor model and Möbius transformation-based models for different cases of circular data. Then, we discuss stereographic projection-based model, rigid-rotation model, geometric model and projective linear transformation-based models for data lying on hyperspheres. In the end, we briefly discuss the kernel-based and diffeomorphism-based nonparametric models. We discuss the inferential properties and computation of the estimates for some of the models.

First Page

333

Last Page

348

DOI

10.1007/978-981-19-1044-9_18

Publication Date

1-1-2022

This document is currently not available here.

Share

COinS