With the increasing urbanization and technological advances, urban driving becomes a complex task. The driver’s mental workload should be able to manage critical situations in challenging driving conditions.

This study focuses on a driver’s physiological changes using portable sensors in different urban routes. Specifically, the Electrodermal Activity, Electromyogram, Heart Rate and Respiration of ten driving experiments in three types of routes are considered: rest area, city, and highway driving.

Several studies consider driver's stress level recognition using such signals. Classically, researchers extract expert-based features from signals and select the most relevant features for stress level recognition.

The contribution is twofold. This work aims to provide a random forest-based method for the selection of functional variables in order to classify the stress level during real-world driving experience. It considers physiological signals as functions and adapt a procedure introduced by Gregorutti et al. 2015. It combines wavelet decompositions to handle functional variables with random forests allowing to build classifiers as well as to rank the variables according to the random forests permutation importance measure. On the applied side, the proposed method provides a "blind" procedure of driver's stress level classification that does not depend on expert knowledge of physiological signals.