PLS classification of functional data

Authors:
Cristian Preda;Gilbert Saporta;Caroline Lévéder
Affiliations:
CERIM - Département de Statistique, Faculté de Médecine, Université de Lille 2, Lille, France 59045;Chaire de Statistique Appliquée, CEDRIC, CNAM, Paris Cedex 03, France 75141;Danone Vitapole, Palaiseau Cedex, France 91767
Venue:
Computational Statistics
Year:
2007

Citing 0
Cited 4

Forecasting binary longitudinal data by a functional PC-ARIMA model

Computational Statistics & Data Analysis
On local times, density estimation and supervised classification from functional data

Journal of Multivariate Analysis
Funclust: A curves clustering method using functional random variables density approximation

Neurocomputing
Model-based clustering for multivariate functional data

Computational Statistics & Data Analysis

Quantified Score

Hi-index	0.00

Visualization

Abstract

Partial least squares (PLS) approach is proposed for linear discriminant analysis (LDA) when predictors are data of functional type (curves). Based on the equivalence between LDA and the multiple linear regression (binary response) and LDA and the canonical correlation analysis (more than two groups), the PLS regression on functional data is used to estimate the discriminant coefficient functions. A simulation study as well as an application to kneading data compare the PLS model results with those given by other methods.