Feature Selection for Functional Data

TL;DR

This paper explores feature selection methods for functional data, emphasizing a flexible blinding procedure that is consistent and effective across classification, regression, and principal component analysis tasks.

Contribution

It introduces a versatile feature selection approach for functional data based on a user-defined set of functions, demonstrating its consistency and practical effectiveness.

Findings

The proposed method is consistent under general assumptions.

It performs well on real data examples.

The blinding procedure offers flexibility in feature definition.

Abstract

In this paper we address the problem of feature selection when the data is functional, we study several statistical procedures including classification, regression and principal components. One advantage of the blinding procedure is that it is very flexible since the features are defined by a set of functions, relevant to the problem being studied, proposed by the user. Our method is consistent under a set of quite general assumptions, and produces good results with the real data examples that we analyze.