Classification supervis\'ee en grande dimension. Application \`a l'agr\'ement de conduite automobile

TL;DR

This paper presents a supervised classification method for high-dimensional functional data, applied to automotive driving assessment, involving wavelet-based preprocessing, dimensionality reduction, and variable selection.

Contribution

It introduces a three-step approach combining wavelet denoising, compression, and CART-based variable selection for high-dimensional functional data classification.

Findings

Effective variable selection in high-dimensional settings

Improved interpretability of driving data models

Robustness to noise through wavelet denoising

Abstract

This work is motivated by a real work problem: objectivization. It consists in explaining the subjective drivability using physical criteria coming from signals measured during experiments. We suggest an approach for the discriminant variables selection trying to take advantage of the functional nature of the data. The porblem is ill-posed, since the number of explanatory variables is hugely greater than the sample size. The strategy proceeds in three steps: a signal preprocessing including wavelet denoising and synchronization, dimensionality reduction by compression using a common wavelet basis, and finally the selection of useful variables using a stepwise strategy involving successive applications of the CART method.