Spherepc: An R Package for Dimension Reduction on a Sphere

Abstract

Dimension reduction is a technique that can compress given data and reduce noise. Recently, a dimension reduction technique on spheres, called spherical principal curves (SPC), has been proposed. SPC fits a curve that passes through the middle of data with a stationary property on spheres. In addition, a study of local principal geodesics (LPG) is considered to identify the complex structure of data. Through the description and implementation of various examples, this paper introduces an R package [spherepc](https://CRAN.R-project.org/package=spherepc) for dimension reduction of data lying on a sphere, including existing methods, SPC and LPG.