Principal Component Analysis of Multivariate Spatial Functional Data

EasyChair Preprint 14324

Abstract

This paper is dedicated to dimension reduction techniques for multivariate spatially indexed functional data. We introduce an innovative method named Spatial Multivariate Funtional Principal Component Analysis (SMFPCA), which stands for principal component analysis for multivariate spatial functional data. Unlike the conventional Multivariate Karhunen-Loève approach, SMFPCA excels at effectively capturing spatial dependencies among multiples functions.
SMFPCA conducts spectral functional component analysis on multivariate spatial data, encompassing data points located within a regular grid. 
The methodological framework and algorithm for SMFPCA have been developed to address the challenges posed by the lack of suitable methods for handling such data. The efficiency of the proposed methodology has been substantiated through comprehensive assessments of its performance using  and simulated datasets and sea-surface temperature, providing valuable insights into the properties of multivariate spatial functional data within a finite sample.

Keyphrases: Functional Principal Component Analysis, Spatial-functional Principal Component Analysis, functional data analysis, multivariate analysis, spectral analysis

@booklet{EasyChair:14324,
  author    = {Idris Si-Ahmed and Leila Hamdad and Christelle Judith Agonkoui and Yoba Kande and Sophie Dabo-Niang},
  title     = {Principal Component Analysis of Multivariate Spatial Functional Data},
  howpublished = {EasyChair Preprint 14324},
  year      = {EasyChair, 2024}}