Abstract:
The Gautrain is a rapid rail system in Gauteng, a province of South Africa. Currently, little is
known about what causes the behaviour of Gautrain passengers concerning their choice of mode
for their first- and last-mile trips. This thesis is a study of the first- and last-mile mode choice
behaviour of Gautrain passengers. The study had three main aims. First, it aimed to develop the
most accurate and statistically significant models of both first- and last-mile behaviour. Part of
this aim was to try to understand the effect of non-traders on the stated preference data. To achieve
this end, models were estimated with and without non-traders in the sample. Second, the study
attempted to determine if there were any differences between commuters’ first- and last-mile
behaviour. For this purpose, a stated preference survey was developed and put online. It was then
marketed on the social media platforms of the Gautrain Management Agency, the body that
manages the rail system. The results of the survey were analysed and cleaned of errors. Different
discrete choice models were derived from the data to find the best fitting model structure. The
last aim was to see how the developed models could be used to plan future access and egress
services. Significant discrete choice models were developed on a reasonably representative
sample of the Gautrain population. The best-fitting model structure was a nested one for both
first- and last-mile trips. These models show that while first- and last-mile commuter behaviour
was similar, there were some differences. In particular, people are much more sensitive to invehicle
time for their last-mile journey in comparison to their first-mile journey. This exercise
showed that although non-trader data reduced the effectiveness and significance of the model
slightly, the data did not change the overall picture shown by the models without non-traders.
