Can technologically assisted travel surveys improve mode choice modelling? Differences and improvements compared to travel conventional survey data.

Abstract

Travel surveys often serve as the primary input for the creation of traffic simulations models, in particular as the
data basis for the estimation of mode choice models. In this paper we investigate the differences of technologically
assisted data (TAD) and conventional travel survey data for mode choice modelling. We describe the necessary
steps to enable the use of TAD for mode choice modelling like data pre-processing, choice set generation and the
adaptation of the travel times in the choice set to fit those in the TAD. Model estimation shows that TAD enables a
better prediction of modal splits compared to the models estimated on the conventional survey data. Furthermore
TAD offers a clear advantage when predicting walking times in the trips. Lastly, the influence of misclassification
in the automated mode detection on the prediction of modal split and walking times is studied. Finally, a short
overview is given on how TAD can be used for route choice modelling which is a further advantage of the new
collection technology.