Multiple Correspondence Analysis of Fuzzyfied Task Performance and Psycho-Physiological Test Data: Use in Real Car Following Situations

Abstract

This article shows the role that Fuzzy Space Windowing (FSW) and Multiple Correspondence Analysis (MCA) may play in the prospect of analyzing two different data sets separately, first and together, afterwards. This methodology is illustrated with a car-following example with 123 individuals. The first data set contains each individual's psychological profile, obtained through tests and questionnaires (23 variables); the second data set contains the signals acquired from real-life driving situations with an instrumented vehicle (6 time variables). The MCA of the first set yields individual typologies, for example, that conscientious drivers declare fewer errors in driving than impulsive and stressed drivers, or that drivers with high cognitive levels have a tendency to declare more violations than other drivers do. The MCA of the second set highlights inter-driver differences (e.g., speed, headway and pedal use). Other factors, such as infrastructure, also have an impact on driving behavior. For example, drivers have a tendency to drive under the speed limit on urban roads and over the speed limit on highways. A third and final MCA is then performed on the most informative variables found with the two previous MCAs.